Upload docs/

.gitattributes

@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
 pairwise_prompts_v2.1.json filter=lfs diff=lfs merge=lfs -text
+docs/paper/figs/fig_encoding_top3.pdf filter=lfs diff=lfs merge=lfs -text

docs/ENCODER_SCRIPTS.md

@@ -0,0 +1,19 @@
+# Per-task encoder scripts
+
+Each **complex** prediction type has its own Python file that reads how predictions are recorded (bbox, polygon, keypoint, etc.) and outputs one image or text file per encoding variant.
+
+| Task | Script | Prediction format | Outputs |
+|------|--------|--------------------|---------|
+| **depth_estimation** | `src/encoders/encode_depth.py` | `predictions[].url` (3 colormap URLs) | 3 images per sample: plasma, turbo, gray |
+| **object_detection** | `src/encoders/encode_object_detection.py` | `predictions_type: "bbox"`, `predictions[].bbox` [x1,y1,x2,y2], `label`, `color_hex` | original, box_only, box_label, text_xyxy.txt, text_xywh.txt |
+| **instance_segmentation** | `src/encoders/encode_instance_segmentation.py` | Polygon: `predictions[].polygon` [[x,y],...]. RLE: `predictions[].rle` (needs pycocotools to decode) | original, enc1..enc6 (overlay images), enc7_json.txt |
+| **semantic_segmentation** | `src/encoders/encode_semantic_segmentation.py` | `predictions[].polygon` [[[x,y],...]] or [[x,y],...], `label` | same 7 as instance (enc1–7) |
+| **referring_segmentation** | `src/encoders/encode_referring_segmentation.py` | `predictions[].polygon`, `label` (referring expression) | original, fill, contour, fill_contour, json.txt |
+| **keypoint** | `src/encoders/encode_keypoint.py` | `predictions[].keypoint` (17×3: x,y,conf per COCO keypoint) | original, enc_a (same color), enc_b (per-person), enc_c (per-limb), enc_d_json.txt |
+| **generation_*** / **lowlevel-*** | `src/encoders/encode_generation_lowlevel.py` | `predictions[].url` or `predictions[].image` (prediction image URL) | 1 image per sample |
+
+**Shared:** `src/common.py` provides `load_image_id_to_url()` (from `images/*.json`) and `fetch_image_cv2(url)` for overlay tasks.
+
+**Requirement for overlay tasks:** `images/*.json` (or root `images.json`) must list each image with `id` and `url` so that `image_id` in annotations can be resolved to the original image URL.
+
+**Run all:** `./run_encoding.sh` (uses 24 threads, 2 samples per task).

docs/ENCODING_VARIANTS.md

@@ -0,0 +1,52 @@
+# Encoding variants per task (from ref/task_encode_req.txt)
+
+Each task must be visualized in **N** ways = full combination of the listed variables.
+
+**Required for overlay encodings:** `images/*.json` (or root `images.json`) must list each image with `id` and `url` (original image URL). The annotation’s `image_id` is used to look up this URL; overlays are drawn on the image fetched from that URL.
+
+## Object Detection
+- **Pixel**: Style {box-only*, box+label}, Mask embedding {overlay*, separate}, Label format {none*, color-description}
+- **Text**: Format {[x1,y1,x2,y2]*, [x,y,w,h]}
+- **Modality**: {image-only*, text-only, image+text}
+- Combinations: 2×2×2×2×3 = **48** (if all independent)
+
+## Instance Segmentation
+- **Pixel**: Sub-sampling {sub-sampled, original}, Opacity α {0.3, 0.5*, 1.0}, Label overlay {none*, text-labels}, Color scheme {color-by-class*, color-by-instance}
+- **Text**: Format {polygon*, RLE}
+- **Modality**: {image-only*, text-only, image+text}
+- Ref implements 7 encodings (enc1–enc7).
+
+## Semantic Segmentation
+- **Pixel**: Sub-sampling, Opacity α {0.3, 0.5*, 1.0}, Label overlay {none*, text-labels}, Color scheme {standard-palette*, random}
+- **Modality**: {image-only*, text-only, image+text}
+
+## Referring Segmentation
+- **Pixel**: Sub-sampling, Opacity α {0.3, 0.5*, 1.0}, Mask style {fill*, contour, fill+contour}
+- **Text**: Format {polygon*, RLE}
+- **Modality**: {image-only*, text-only}
+
+## Keypoint
+- **Pixel**: Style {points-only, skeleton*}, Color scheme {same-color, color-by-instance*, color-by-part}
+- **Text**: Format {flat-list, part-keyed-JSON*, COCO-style}
+- **Modality**: {image-only*, text-only, best-visual+text}
+- Ref implements 4 encodings (a,b,c,d).
+
+## Depth
+- **Pixel**: Colormap {plasma*, viridis, turbo} — **3 ways** (pre-rendered; just download).
+- **Modality**: image-only.
+
+## Low-level / Generation
+- No encoding variants (edited/generated images only).
+
+
+task – always set to the task name.
+pixel_style – e.g. box_only, box_label, points_only, skeleton (only for object_detection and keypoint).
+mask_embed – overlay or separate (segmentation tasks).
+label_fmt – none, color_filled_rect, text_only (object_detection only).
+sub_sampling – original or sub_sampled (instance/semantic/referring).
+opacity – 0.3, 0.5, 1.0 (segmentation).
+label_overlay – none or text_labels (instance/semantic).
+color_scheme – e.g. color_by_class, standard_palette, same_color (instance/semantic/keypoint).
+mask_style – fill, contour, fill_contour (referring only).
+text_format – e.g. xyxy, polygon, flat_list (per-task text formats).
+colormap – plasma, turbo, gray (depth only).

docs/ENCODING_VARIANT_COUNTS.md

@@ -0,0 +1,34 @@
+# Full encoding variant counts (ref/task_encode_req.txt)
+
+**Current output: image-only + text-only** (image+text combinations not generated yet).
+
+## Object Detection
+- **Pixel:** Style(2) × Mask(2) × Label format(3) = **12** image-only
+- **Text:** Format(2) = **2** text-only  
+- **Output:** 12 + 2 = **14** per sample ✅
+
+## Instance Segmentation
+- **Pixel:** Sub-sampling(2) × Opacity(3) × Label overlay(2) × Color scheme(2) = **24**
+- **Text:** Format(2): polygon, RLE
+- **Output:** 24 + 2 = **26** per sample ✅
+
+## Semantic Segmentation
+- **Pixel:** Sub-sampling(2) × Opacity(3) × Label overlay(2) × Color scheme(2) = **24**
+- **Text:** Format(2): polygon, RLE
+- **Output:** 24 + 2 = **26** per sample ✅
+
+## Referring Segmentation
+- **Pixel:** Sub-sampling(2) × Opacity(3) × Mask style(3) = **18**
+- **Text:** Format(2): polygon, RLE
+- **Output:** 18 + 2 = **20** per sample ✅
+
+## Keypoint
+- **Pixel:** Style(2): points-only, skeleton × Color scheme(3) = **6**
+- **Text:** Format(3): flat-list, part-keyed-JSON, COCO-style = **3**
+- **Output:** 6 + 3 = **9** per sample ✅
+
+## Depth
+- **Pixel:** Colormap(3). **Modality:** image-only → **3** ✅
+
+## Low-level / Generation
+- No variations (single output type each).

docs/PIPELINE_OVERVIEW.md

@@ -0,0 +1,116 @@
+# Pipeline Overview: `run_encoding.sh` → `run_pairwise_prompts_vfinal.sh`
+
+## What each script does
+
+### `run_encoding.sh` — Step 1: Encode annotations into visual/text representations
+
+Converts raw annotations into per-annotation encoded outputs (images or text strings) that can be embedded into LLM prompts.
+
+**Inputs**
+- `annotations/` — task annotation JSONs (e.g. `object_detection.json`, `keypoint.json`, …)
+- `images/` — image metadata JSONs mapping `image_id` → URL/path
+
+**Steps**
+1. **Inventory** (`src/preprocess/inventory_annotations.py`) — scans `annotations/` and logs unique `(task, prediction_type, prediction_keys)` combinations.
+2. **Encode** — runs one encoder per task in parallel (up to `NUM_THREADS=24` threads each):
+   - `encode_depth.py`
+   - `encode_object_detection.py`
+   - `encode_instance_segmentation.py`
+   - `encode_semantic_segmentation.py`
+   - `encode_referring_segmentation.py`
+   - `encode_keypoint.py`
+   - *(generation / low-level tasks disabled but present as commented code)*
+3. **Verify** (`src/verify/verify_and_visualize_variants.py`) — checks variant consistency and produces `output/encoding_variants_visual.png`.
+
+**Outputs** → `output/encoded/<task>/` — one encoded file (image or text) per annotation per encoding variant.
+
+**Key flags**
+| Variable | Default | Effect |
+|---|---|---|
+| `SAMPLES` | `2` | Number of `(image_id, instance_type, synthetic?)` combos to process; `0` = all |
+| `SKIP_SUBSAMPLING` | `0` | Skip instance/semantic/referring segmentation encoders |
+| `BASE_ONLY` | `0` | Only produce `is_base==1` encodings |
+| `BASE_ABLATION` | `0` | Produce base + 1-variant ablation set |
+
+---
+
+### `run_pairwise_prompts_vfinal.sh` — Step 2: Build pairwise and scoring prompt JSON
+
+Samples pairs of model predictions from the encoded pool and wraps them into structured LLM judge prompts (pairwise comparison + scoring).
+
+**Inputs**
+- `output/encoded/` — outputs from Step 1 (or pass `--skip-encode` to reuse existing)
+- `docs/encoding_variants_summary_vfinal3-sem-enc.csv` — CSV listing which encoding variant IDs are `is_final==1`; controls which encodings are used
+- `annotations/` (via the same `ANNOTATIONS_DIR` path) — annotation JSONs for grouping and sampling
+- Original images (fetched on demand from URL if not cached locally)
+
+**Steps**
+1. **Encode (inline)** — unless `--skip-encode` is passed, re-runs encoders limited to `is_final==1` variants from the CSV.
+2. **Pairwise prompts** (`src/pairwise/build_pairwise_prompts_vfinal.py`) — for each task, samples up to `IMAGES_PER_TASK` unique image IDs and `PAIRS_PER_TASK=40` annotation pairs, then builds a structured prompt JSON with images and instructions.
+3. **Scoring prompts** (`src/scoring/build_scoring_prompts.py`) — samples `SCORING_COMBOS_PER_TASK=20` `(image_id, coi, error_type, prompt)` combos × `SCORING_ANNOTATIONS_PER_COMBO=5` annotations each, appended to the same JSON with `question_type="scoring"`.
+4. **Convert messages** (`src/tools/convert_messages.py`) — transforms the raw prompt JSON into a `_messages.json` format ready to send to an LLM API.
+
+**Outputs**
+- `output/pairwise/pairwise_prompts_vFinal3-sem-enc.json`
+- `output/pairwise/pairwise_prompts_vFinal3-sem-enc_messages.json` ← primary deliverable
+
+---
+
+## Can they be a single pipeline driven by `data_json_v2` + `images_v2`?
+
+**Yes — with environment variable overrides.** The codebase is already wired for this:
+
+- `src/common.py` reads `$JEP_ANNOTATIONS_DIR` (default: `annotations/`) and `$JEP_IMAGES_DIR` (default: `images/`) for all encoders.
+- `src/tools/reencode_and_push_annotation_rank_to_hub.py` shows the pattern in use: it sets `JEP_ANNOTATIONS_DIR` and `JEP_IMAGES_DIR` before calling encoders.
+
+**However**, two modules still hard-code `ROOT / "annotations"` and are **not** yet env-aware:
+- `src/preprocess/inventory_annotations.py` (line 15)
+- `src/pairwise/build_pairwise_prompts.py` (line 40)
+
+### Minimal changes needed to support `data_json_v2` / `images_v2`
+
+1. **`src/preprocess/inventory_annotations.py`** — replace hard-coded path:
+   ```python
+   # Before
+   ANNOTATIONS_DIR = ROOT / "annotations"
+   # After
+   ANNOTATIONS_DIR = Path(os.environ.get("JEP_ANNOTATIONS_DIR", str(ROOT / "annotations")))
+   ```
+
+2. **`src/pairwise/build_pairwise_prompts.py`** — same fix at line 40 (and optionally `IMAGES_DIR`).
+
+3. **Run the pipeline** with env overrides:
+   ```bash
+   export JEP_ANNOTATIONS_DIR=/raid/icy/judge-encode-prompt/data_json_v2
+   export JEP_IMAGES_DIR=/raid/icy/judge-encode-prompt/images_v2
+   bash run_encoding.sh
+   bash run_pairwise_prompts_vfinal.sh
+   ```
+
+No other changes are required — all encoders inherit the env vars via `src/common.py`.
+
+---
+
+## End-to-end data flow
+
+```
+data_json_v2/           images_v2/
+  *.json (annotations)    *.json (image metadata: image_id → URL/path)
+        │                       │
+        └──────────┬────────────┘
+                   ▼
+         run_encoding.sh
+           inventory_annotations.py  →  logs task inventory
+           encode_<task>.py (×6, parallel)  →  output/encoded/<task>/
+           verify_and_visualize_variants.py
+                   │
+                   ▼
+    run_pairwise_prompts_vfinal.sh
+           build_pairwise_prompts_vfinal.py  →  pairwise prompt JSON
+           build_scoring_prompts.py          →  scoring prompts appended
+           convert_messages.py
+                   │
+                   ▼
+    output/pairwise/pairwise_prompts_vFinal3-sem-enc_messages.json
+    (LLM-ready judge prompts)
+```

docs/TASK_REPRESENTATIONS.md

@@ -0,0 +1,95 @@
+# Task & prediction type inventory
+
+Generated by `src/preprocess/inventory_annotations.py`. Use this to index what representation types exist per CV task.
+
+## Unique values across all annotations
+
+### Tasks
+```
+depth_estimation generation_controllable generation_editing generation_inpainting_high_level generation_inpainting_low_level generation_t2i instance_segmentation keypoint lowlevel-deblur lowlevel-derain lowlevel-desnow lowlevel-super-resolution object_detection referring_segmentation semantic_segmentation
+```
+
+### Predictions type (field: predictions_type / prediction_type)
+```
+bbox image keypoint mask npy polygon rle
+```
+
+### Prediction-item keys (keys inside each element of `predictions[]`) by task
+
+- **depth_estimation**: `image, label, npy, url`
+- **generation_controllable**: `image, url`
+- **generation_editing**: `image, url`
+- **generation_inpainting_high_level**: `image, url`
+- **generation_inpainting_low_level**: `image, url`
+- **generation_t2i**: `image, url`
+- **instance_segmentation**: `bbox, class_iou, instance_type, label, matched_gt_iscrowd, matrix, other_metadata, polygon, rle, score`
+- **keypoint**: `error_type, keypoint, label, score`
+- **lowlevel-deblur**: `image, url`
+- **lowlevel-derain**: `image, url`
+- **lowlevel-desnow**: `image, url`
+- **lowlevel-super-resolution**: `image, url`
+- **object_detection**: `bbox, class_id, color_hex, file_path, label`
+- **referring_segmentation**: `label, matrix, polygon`
+- **semantic_segmentation**: `class_id, class_iou, color_hex, image, label, matrix, polygon, url`
+
+## Per-file summary
+
+### depth_annotations.json
+- Tasks: `['depth_estimation']`
+- predictions_type: `['npy']`
+- prediction item keys: `['image', 'label', 'npy', 'url']`
+- annotation count: 358
+
+### generation_annotations.json
+- Tasks: `['generation_controllable', 'generation_editing', 'generation_inpainting_high_level', 'generation_inpainting_low_level', 'generation_t2i']`
+- predictions_type: `['image']`
+- prediction item keys: `['image', 'url']`
+- annotation count: 248
+
+### instance_segmentation_model_annotations.json
+- Tasks: `['instance_segmentation']`
+- predictions_type: `['rle']`
+- prediction item keys: `['bbox', 'class_iou', 'instance_type', 'label', 'matched_gt_iscrowd', 'matrix', 'other_metadata', 'rle', 'score']`
+- annotation count: 362
+
+### instance_segmentation_synthetic_annotations.json
+- Tasks: `['instance_segmentation']`
+- predictions_type: `['polygon']`
+- prediction item keys: `['label', 'matrix', 'polygon']`
+- annotation count: 2021
+
+### keypoint_annotations.json
+- Tasks: `['keypoint']`
+- predictions_type: `['keypoint']`
+- prediction item keys: `['error_type', 'keypoint', 'label', 'score']`
+- annotation count: 366
+
+### lowlevel_annotations.json
+- Tasks: `['lowlevel-deblur', 'lowlevel-derain', 'lowlevel-desnow', 'lowlevel-super-resolution']`
+- predictions_type: `['image']`
+- prediction item keys: `['image', 'url']`
+- annotation count: 3005
+
+### object_detection_annotations.json
+- Tasks: `['object_detection']`
+- predictions_type: `['bbox']`
+- prediction item keys: `['bbox', 'class_id', 'color_hex', 'file_path', 'label']`
+- annotation count: 1164
+
+### referring_segmentation_model_annotations.json
+- Tasks: `['referring_segmentation']`
+- predictions_type: `['polygon']`
+- prediction item keys: `['label', 'matrix', 'polygon']`
+- annotation count: 500
+
+### referring_segmentation_synthetic_annotations.json
+- Tasks: `['referring_segmentation']`
+- predictions_type: `['polygon']`
+- prediction item keys: `['label', 'matrix', 'polygon']`
+- annotation count: 2017
+
+### semantic_segmentation_annotations.json
+- Tasks: `['semantic_segmentation']`
+- predictions_type: `['mask', 'polygon']`
+- prediction item keys: `['class_id', 'class_iou', 'color_hex', 'image', 'label', 'matrix', 'polygon', 'url']`
+- annotation count: 936

docs/analysis_model_task.csv

@@ -0,0 +1,31 @@
+model,task,accuracy,correct,total
+Qwen2.5-VL-3B-Instruct-judge-enc,depth_estimation,0.6118,208,340
+Qwen2.5-VL-3B-Instruct-judge-enc,generation_controllable,0.5816,196,337
+Qwen2.5-VL-3B-Instruct-judge-enc,generation_editing,0.55,187,340
+Qwen2.5-VL-3B-Instruct-judge-enc,generation_inpainting_high_level,0.6147,209,340
+Qwen2.5-VL-3B-Instruct-judge-enc,generation_inpainting_low_level,0.6029,205,340
+Qwen2.5-VL-3B-Instruct-judge-enc,generation_t2i,0.6912,235,340
+Qwen2.5-VL-3B-Instruct-judge-enc,instance_segmentation,0.6161,626,1016
+Qwen2.5-VL-3B-Instruct-judge-enc,keypoint,0.6529,222,340
+Qwen2.5-VL-3B-Instruct-judge-enc,lowlevel-deblur,0.7818,172,220
+Qwen2.5-VL-3B-Instruct-judge-enc,lowlevel-derain,0.8765,298,340
+Qwen2.5-VL-3B-Instruct-judge-enc,lowlevel-desnow,0.9149,301,329
+Qwen2.5-VL-3B-Instruct-judge-enc,lowlevel-super-resolution,0.5441,185,340
+Qwen2.5-VL-3B-Instruct-judge-enc,object_detection,0.7422,757,1020
+Qwen2.5-VL-3B-Instruct-judge-enc,referring_segmentation,0.5833,595,1020
+Qwen2.5-VL-3B-Instruct-judge-enc,semantic_segmentation,0.6079,727,1196
+Qwen2.5-VL-7B-Instruct-judge-enc,depth_estimation,0.7176,244,340
+Qwen2.5-VL-7B-Instruct-judge-enc,generation_controllable,0.6103,166,272
+Qwen2.5-VL-7B-Instruct-judge-enc,generation_editing,0.5605,190,339
+Qwen2.5-VL-7B-Instruct-judge-enc,generation_inpainting_high_level,0.6206,211,340
+Qwen2.5-VL-7B-Instruct-judge-enc,generation_inpainting_low_level,0.5941,202,340
+Qwen2.5-VL-7B-Instruct-judge-enc,generation_t2i,0.6735,229,340
+Qwen2.5-VL-7B-Instruct-judge-enc,instance_segmentation,0.6248,631,1010
+Qwen2.5-VL-7B-Instruct-judge-enc,keypoint,0.6706,228,340
+Qwen2.5-VL-7B-Instruct-judge-enc,lowlevel-deblur,0.7318,161,220
+Qwen2.5-VL-7B-Instruct-judge-enc,lowlevel-derain,0.9059,308,340
+Qwen2.5-VL-7B-Instruct-judge-enc,lowlevel-desnow,0.9331,307,329
+Qwen2.5-VL-7B-Instruct-judge-enc,lowlevel-super-resolution,0.5971,203,340
+Qwen2.5-VL-7B-Instruct-judge-enc,object_detection,0.7529,768,1020
+Qwen2.5-VL-7B-Instruct-judge-enc,referring_segmentation,0.6206,633,1020
+Qwen2.5-VL-7B-Instruct-judge-enc,semantic_segmentation,0.6279,751,1196

docs/analysis_pairwise_encoding_by_task.csv

@@ -0,0 +1,30 @@
+task,encoding,encoding_id,accuracy,correct,total
+depth_estimation,plasma,97,0.6647,452,680
+generation_controllable,generation_controllable,,0.5944,362,609
+generation_editing,generation_editing,,0.5552,377,679
+generation_inpainting_high_level,generation_inpainting_high_level,,0.6176,420,680
+generation_inpainting_low_level,generation_inpainting_low_level,,0.5985,407,680
+generation_t2i,generation_t2i,,0.6824,464,680
+instance_segmentation,1742,1742,0.5893,396,672
+instance_segmentation,pixel_ss1_m0_o0_l1_c1_b1,17,0.6391,425,665
+instance_segmentation,text_polygon,42,0.6328,436,689
+keypoint,pixel_s1_c1_m0,89,0.6618,450,680
+lowlevel-deblur,lowlevel-deblur,,0.7568,333,440
+lowlevel-derain,lowlevel-derain,,0.8912,606,680
+lowlevel-desnow,lowlevel-desnow,,0.924,608,658
+lowlevel-super-resolution,lowlevel-super-resolution,,0.5706,388,680
+object_detection,0305,305,0.7662,521,680
+object_detection,pixel_s1_m0,3,0.6765,460,680
+object_detection,text_xyxy,5,0.8,544,680
+referring_segmentation,7080,7080,0.6029,410,680
+referring_segmentation,pixel_ss1_m0_o0_m2,70,0.5809,395,680
+referring_segmentation,text_polygon,80,0.6221,423,680
+semantic_segmentation,4649,4649,0.6288,454,722
+semantic_segmentation,pixel_ss0_m0,44,0.3333,2,6
+semantic_segmentation,pixel_ss0_m1,45,0.5,3,6
+semantic_segmentation,pixel_ss1_m0_o0_l1_c0,49,0.608,439,722
+semantic_segmentation,pixel_ss1_m0_o0_l1_c1,50,0.5625,27,48
+semantic_segmentation,pixel_ss1_m0_o1_l1_c0,53,0.5625,27,48
+semantic_segmentation,pixel_ss1_m1_o0_l1_c0,57,0.6429,45,70
+semantic_segmentation,text_matrix,46,0.6274,453,722
+semantic_segmentation,text_polygon,63,0.5833,28,48

docs/analysis_pairwise_overall_by_task.csv

@@ -0,0 +1,16 @@
+task,accuracy,correct,total
+depth_estimation,0.6647,452,680
+generation_controllable,0.5944,362,609
+generation_editing,0.5552,377,679
+generation_inpainting_high_level,0.6176,420,680
+generation_inpainting_low_level,0.5985,407,680
+generation_t2i,0.6824,464,680
+instance_segmentation,0.6204,1257,2026
+keypoint,0.6618,450,680
+lowlevel-deblur,0.7568,333,440
+lowlevel-derain,0.8912,606,680
+lowlevel-desnow,0.924,608,658
+lowlevel-super-resolution,0.5706,388,680
+object_detection,0.7475,1525,2040
+referring_segmentation,0.602,1228,2040
+semantic_segmentation,0.6179,1478,2392

docs/analysis_ranking_by_task.csv

@@ -0,0 +1,7 @@
+task,accuracy_exact,accuracy_top1,correct_exact,correct_top1,total
+depth_estimation,0.3281,0.4219,21,27,64
+instance_segmentation,0.3478,0.5,48,69,138
+keypoint,0.0,0.1935,0,6,31
+object_detection,0.4569,0.5259,53,61,116
+referring_segmentation,0.1983,0.3554,24,43,121
+semantic_segmentation,0.0324,0.1871,9,52,278

docs/analysis_scoring_by_task.csv

@@ -0,0 +1,7 @@
+task,accuracy,correct,total
+depth_estimation,0.3,72,240
+instance_segmentation,0.3463,169,488
+keypoint,0.1211,23,190
+object_detection,0.1376,71,516
+referring_segmentation,0.1343,87,648
+semantic_segmentation,0.2744,422,1538

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_10pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,331,128
+generation_controllable,396,39
+generation_editing,400,40
+generation_inpainting_high_level,400,50
+generation_inpainting_low_level,400,80
+generation_t2i,400,400
+instance_segmentation,283,35
+keypoint,260,82
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,341,40
+object_detection,345,38
+referring_segmentation,349,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_15pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,331,128
+generation_controllable,207,38
+generation_editing,345,40
+generation_inpainting_high_level,400,50
+generation_inpainting_low_level,400,80
+generation_t2i,400,400
+instance_segmentation,283,35
+keypoint,260,82
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,345,38
+referring_segmentation,349,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_20pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,331,128
+generation_controllable,207,38
+generation_editing,242,40
+generation_inpainting_high_level,232,50
+generation_inpainting_low_level,400,80
+generation_t2i,400,400
+instance_segmentation,283,35
+keypoint,260,82
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,345,38
+referring_segmentation,349,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_25pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,331,128
+generation_controllable,207,38
+generation_editing,242,40
+generation_inpainting_high_level,217,50
+generation_inpainting_low_level,304,79
+generation_t2i,272,272
+instance_segmentation,278,34
+keypoint,260,82
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,319,38
+referring_segmentation,349,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_30pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,331,128
+generation_controllable,207,38
+generation_editing,242,40
+generation_inpainting_high_level,217,50
+generation_inpainting_low_level,304,79
+generation_t2i,272,272
+instance_segmentation,207,30
+keypoint,260,82
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,248,35
+referring_segmentation,221,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_35pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,178,100
+generation_controllable,207,38
+generation_editing,242,40
+generation_inpainting_high_level,217,50
+generation_inpainting_low_level,304,79
+generation_t2i,272,272
+instance_segmentation,193,29
+keypoint,190,75
+lowlevel-deblur,249,40
+lowlevel-derain,368,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,247,35
+referring_segmentation,188,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_40pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,121,83
+generation_controllable,207,38
+generation_editing,242,40
+generation_inpainting_high_level,217,50
+generation_inpainting_low_level,304,79
+generation_t2i,272,272
+instance_segmentation,193,29
+keypoint,190,75
+lowlevel-deblur,113,40
+lowlevel-derain,291,40
+lowlevel-desnow,344,60
+lowlevel-super-resolution,315,40
+object_detection,247,35
+referring_segmentation,188,39

docs/encoding_variants_drop_bottom_accuracy/drop_bottom_5pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id
+depth_estimation,400,138
+generation_controllable,396,39
+generation_editing,400,40
+generation_inpainting_high_level,400,50
+generation_inpainting_low_level,400,80
+generation_t2i,400,400
+instance_segmentation,283,35
+keypoint,318,84
+lowlevel-deblur,258,40
+lowlevel-derain,400,40
+lowlevel-desnow,387,60
+lowlevel-super-resolution,400,40
+object_detection,345,38
+referring_segmentation,349,39

docs/encoding_variants_summary.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem
+1,0,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,-,-,-
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,
+3,1,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,
+4,0,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,
+5,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,
+6,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,
+7,0,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+8,0,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+9,0,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,
+11,0,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+14,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+15,0,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+17,1,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+23,0,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+31,0,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+42,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+43,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+44,0,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+45,0,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+46,1,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+47,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,
+49,1,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+50,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,
+53,0,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,
+57,0,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,
+63,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+64,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+65,0,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+66,0,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+67,0,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+68,0,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,
+69,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,
+70,1,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,
+71,0,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,
+74,0,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,
+80,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+81,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,
+83,0,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+88,0,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,
+89,1,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+90,0,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,
+92,0,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+94,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,
+95,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,
+96,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,
+97,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,
+98,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,
+99,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,
+100,1,1,1,low_level,-,-,-,-,-,-,-,-,-,-,-,,,
+101,1,1,1,generation,-,-,-,-,-,-,-,-,-,-,-,,,
+1742,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742
+305,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305
+7080,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080
+4649,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649

docs/encoding_variants_summary_results.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem,avg_acc,InternVL3-1B-Instruct-judge-enc,InternVL3-2B-Instruct-judge-enc,InternVL3-38B-Instruct-judge-enc,InternVL3-78B-Instruct-judge-enc,InternVL3-8B-Instruct-judge-enc,Qwen2.5-VL-32B-Instruct-judge-enc,Qwen2.5-VL-3B-Instruct-judge-enc,Qwen2.5-VL-72B-Instruct-API-judge-enc,Qwen2.5-VL-7B-Instruct-judge-enc,gemma-3-12b-it-judge-enc,gemma-3-4b-it-judge-enc
+1,0,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,-,-,-,,,,,,,,,,,,
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+3,1,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,,0.5915,0.4889 (incomplete),0.5059 (incomplete),0.6780 (incomplete),0.6441 (incomplete),0.5354 (incomplete),0.6650 (incomplete),0.5694,0.6522 (incomplete),0.5812,0.5981 (incomplete),0.5984 (incomplete)
+4,0,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+5,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,,0.6402,0.5314 (incomplete),0.6028 (incomplete),0.6644 (incomplete),0.6485 (incomplete),0.6064 (incomplete),0.7337 (incomplete),0.6801,0.8270 (incomplete),0.6809,0.5547 (incomplete),0.5416 (incomplete)
+6,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,,,,,,,,,,,,,
+7,0,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+8,0,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+9,0,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+11,0,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+14,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+15,0,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+17,1,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,,0.5652,0.4758 (incomplete),0.5197 (incomplete),0.6273 (incomplete),0.5263 (incomplete),0.5236 (incomplete),0.6012 (incomplete),0.5409,0.5932 (incomplete),0.5835,0.6133 (incomplete),0.6250 (incomplete)
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+23,0,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+31,0,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,,,,,,,,,,,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,,,,,,,,,,,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,,,,,,,,,,,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,,,,,,,,,,,,,
+42,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,,0.5671,0.5288 (incomplete),0.5524 (incomplete),0.5653 (incomplete),0.6290 (incomplete),0.5568 (incomplete),0.6645 (incomplete),0.5321,0.6656 (incomplete),0.5582,0.5513 (incomplete),0.5237 (incomplete)
+43,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,,,,,,,,,,,,,
+44,0,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+45,0,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+46,1,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+47,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,,,,,,,,,,,,,
+49,1,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+50,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,,,,,,,,,,,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,,,,,,,,,,,,,
+53,0,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,,,,,,,,,,,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,,,,,,,,,,,,,
+57,0,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,,,,,,,,,,,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,,,,,,,,,,,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,,,,,,,,,,,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,,,,,,,,,,,,,
+63,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,,,,,,,,,,,,,
+64,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,,,,,,,,,,,,,
+65,0,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+66,0,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+67,0,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,,,,,,,,,,,,,
+68,0,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,,,,,,,,,,,,,
+69,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,,,,,,,,,,,,,
+70,1,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,,0.5459,0.4864 (incomplete),0.4828 (incomplete),0.6569 (incomplete),,0.4988 (incomplete),0.6475 (incomplete),0.4668,0.6192 (incomplete),0.5437,0.5862 (incomplete),0.5012 (incomplete)
+71,0,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,,,,,,,,,,,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,,,,,,,,,,,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,,,,,,,,,,,,,
+74,0,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,,,,,,,,,,,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,,,,,,,,,,,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,,,,,,,,,,,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,,,,,,,,,,,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,,,,,,,,,,,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,,,,,,,,,,,,,
+80,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,,0.5575,0.5123 (incomplete),0.5401 (incomplete),0.5850 (incomplete),,0.5205 (incomplete),0.6425 (incomplete),0.5257,0.6452 (incomplete),0.5437,0.5726 (incomplete),0.5036 (incomplete)
+81,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,,,,,,,,,,,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,,,,,,,,,,,,,
+83,0,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,,,,,,,,,,,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,,,,,,,,,,,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,,,,,,,,,,,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,,,,,,,,,,,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,,,,,,,,,,,,,
+88,0,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,,,,,,,,,,,,,
+89,1,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,,0.5189,0.3980 (incomplete),0.5409 (incomplete),0.4909 (incomplete),,0.4732 (incomplete),0.5875 (incomplete),0.5533,0.5588 (incomplete),0.5631,0.5980 (incomplete),0.4829 (incomplete)
+90,0,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,,,,,,,,,,,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,,,,,,,,,,,,,
+92,0,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,,,,,,,,,,,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,,,,,,,,,,,,,
+94,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,,,,,,,,,,,,,
+95,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,,,,,,,,,,,,,
+96,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,,,,,,,,,,,,,
+97,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,,0.5451,0.5136 (incomplete),0.5239 (incomplete),0.5905 (incomplete),,0.5514 (incomplete),0.4925 (incomplete),0.5144,0.5947 (incomplete),0.6172,0.5962 (incomplete),0.4952 (incomplete)
+98,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,,,,,,,,,,,,,
+99,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,,,,,,,,,,,,,
+100,1,1,1,low_level,-,-,-,-,-,-,-,-,-,-,-,,,,0.6587,,,,,,,,,,,
+101,1,1,1,generation,-,-,-,-,-,-,-,-,-,-,-,,,,0.5211,,,,,,,,,,,
+1742,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742,0.5605,0.4987 (incomplete),0.4729 (incomplete),0.6382 (incomplete),0.6667 (incomplete),0.5974 (incomplete),0.6502 (incomplete),0.5271,0.6781 (incomplete),0.4926,0.6131 (incomplete),0.5173 (incomplete)
+305,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305,0.6202,0.5134 (incomplete),0.5332 (incomplete),0.6739 (incomplete),0.6060 (incomplete),0.5485 (incomplete),0.7125 (incomplete),0.6376,0.7694 (incomplete),0.6549,0.6275 (incomplete),0.5428 (incomplete)
+7080,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080,0.5455,0.4840 (incomplete),0.4939 (incomplete),0.6261 (incomplete),,0.4927 (incomplete),0.6500 (incomplete),0.4963,0.6716 (incomplete),0.5196,0.4779 (incomplete),0.5097 (incomplete)
+4649,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649,,,,,,,,,,,,

docs/encoding_variants_summary_v2.1.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem
+1,1,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,-,-,-
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,
+3,1,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,
+4,1,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,
+5,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,
+6,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,
+7,1,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+8,1,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+9,1,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,
+11,1,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+14,1,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+15,1,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+17,1,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+23,1,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+31,1,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+42,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+43,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+44,1,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+45,1,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+46,1,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+47,1,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,
+49,1,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+50,1,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,
+53,1,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,
+57,1,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,
+63,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+64,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+65,1,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+66,1,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+67,1,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+68,1,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,
+69,1,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,
+70,1,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,
+71,1,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,
+74,1,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,
+80,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+81,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,
+83,1,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+88,1,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,
+89,1,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+90,1,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,
+92,1,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+94,1,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,
+95,1,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,
+96,1,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,
+97,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,
+98,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,
+99,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,
+100,1,1,0,low_level,-,-,-,-,-,-,-,-,-,-,-,,,
+101,1,1,0,generation,-,-,-,-,-,-,-,-,-,-,-,,,
+1742,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742
+305,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305
+7080,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080
+4649,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649

docs/encoding_variants_summary_vfinal1.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem
+1,0,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,,,-
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,
+3,1,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,
+4,0,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,
+5,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,
+6,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,
+7,0,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+8,0,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+9,0,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,
+11,0,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+14,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+15,0,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+17,1,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+23,0,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+31,0,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+42,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+43,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+44,0,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+45,0,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+46,0,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+47,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,
+49,0,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+50,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,
+53,0,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,
+57,0,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,
+63,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+64,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+65,0,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+66,0,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+67,0,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+68,0,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,
+69,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,
+70,1,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,
+71,0,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,
+74,0,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,
+80,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+81,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,
+83,0,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+88,0,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,
+89,1,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+90,0,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,
+92,0,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+94,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,
+95,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,
+96,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,
+97,1,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,
+98,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,
+99,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,
+100,1,1,1,low_level,-,-,-,-,-,-,-,-,-,-,-,,,
+101,1,1,1,generation,-,-,-,-,-,-,-,-,-,-,-,,,
+1742,1,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742
+305,1,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305
+7080,1,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080
+4649,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649

docs/encoding_variants_summary_vfinal2.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem
+1,0,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,-,-,-
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,
+3,0,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,
+4,0,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,
+5,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,
+6,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,
+7,0,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+8,0,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+9,0,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,
+11,0,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+14,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+15,0,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+17,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+23,0,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+31,0,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+42,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+43,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+44,0,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+45,0,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+46,1,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+47,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,
+49,1,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+50,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,
+53,0,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,
+57,0,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,
+63,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+64,0,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+65,0,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+66,0,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+67,0,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+68,0,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,
+69,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,
+70,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,
+71,0,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,
+74,0,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,
+80,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+81,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,
+83,0,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+88,0,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,
+89,0,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+90,0,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,
+92,0,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+94,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,
+95,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,
+96,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,
+97,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,
+98,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,
+99,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,
+100,0,1,1,low_level,-,-,-,-,-,-,-,-,-,-,-,,,
+101,0,1,1,generation,-,-,-,-,-,-,-,-,-,-,-,,,
+1742,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742
+305,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305
+7080,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080
+4649,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649

docs/encoding_variants_summary_vfinal3-sem-enc.csv

@@ -0,0 +1,106 @@
+encoding_id,is_final,is_base,is_base_ablation,task,pixel_style,mask_embed,label_fmt,sub_sampling,opacity,label_overlay,color_scheme,instance_bbox,mask_style,text_format,colormap,combo_text_id,combo_image_id,combo_stem
+1,0,0,1,object_detection,box_only,overlay,-,-,-,-,-,-,-,-,-,-,-,-
+2,0,0,0,object_detection,box_only,separate,-,-,-,-,-,-,-,-,-,,,
+3,0,1,1,object_detection,box_label,overlay,-,-,-,-,-,-,-,-,-,,,
+4,0,0,1,object_detection,box_label,separate,-,-,-,-,-,-,-,-,-,,,
+5,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xyxy,-,,,
+6,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,xywh,-,,,
+7,0,0,1,instance_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+8,0,0,1,instance_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+9,0,0,1,instance_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+10,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,none,-,-,-,,,
+11,0,0,1,instance_segmentation,-,overlay,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+12,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+13,0,0,0,instance_segmentation,-,overlay,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+14,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+15,0,1,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+16,0,0,0,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+17,0,0,1,instance_segmentation,-,overlay,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+18,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,none,-,-,-,,,
+19,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_class,dashed,-,-,-,,,
+20,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,none,-,-,-,,,
+21,0,0,0,instance_segmentation,-,overlay,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+22,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+23,0,0,1,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+24,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+25,0,0,0,instance_segmentation,-,overlay,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+26,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,none,-,-,-,,,
+27,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_class,dashed,-,-,-,,,
+28,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,none,-,-,-,,,
+29,0,0,0,instance_segmentation,-,separate,-,original,0.5,none,color_by_instance,dashed,-,-,-,,,
+30,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,none,-,-,-,,,
+31,0,0,1,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_class,dashed,-,-,-,,,
+32,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,none,-,-,-,,,
+33,0,0,0,instance_segmentation,-,separate,-,original,0.5,text_labels,color_by_instance,dashed,-,-,-,,,
+34,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,none,-,-,-,,,
+35,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_class,dashed,-,-,-,,,
+36,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,none,-,-,-,,,
+37,0,0,0,instance_segmentation,-,separate,-,original,1,none,color_by_instance,dashed,-,-,-,,,
+38,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,none,-,-,-,,,
+39,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_class,dashed,-,-,-,,,
+40,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,none,-,-,-,,,
+41,0,0,0,instance_segmentation,-,separate,-,original,1,text_labels,color_by_instance,dashed,-,-,-,,,
+42,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+43,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+44,1,0,1,semantic_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+45,1,0,1,semantic_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+46,1,0,1,semantic_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+47,0,0,1,semantic_segmentation,-,overlay,-,original,0.5,none,standard_palette,-,-,-,-,,,
+48,0,0,0,semantic_segmentation,-,overlay,-,original,0.5,none,random,-,-,-,-,,,
+49,1,1,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+50,1,0,1,semantic_segmentation,-,overlay,-,original,0.5,text_labels,random,-,-,-,-,,,
+51,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,standard_palette,-,-,-,-,,,
+52,0,0,0,semantic_segmentation,-,overlay,-,original,1,none,random,-,-,-,-,,,
+53,1,0,1,semantic_segmentation,-,overlay,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+54,0,0,0,semantic_segmentation,-,overlay,-,original,1,text_labels,random,-,-,-,-,,,
+55,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,standard_palette,-,-,-,-,,,
+56,0,0,0,semantic_segmentation,-,separate,-,original,0.5,none,random,-,-,-,-,,,
+57,1,0,1,semantic_segmentation,-,separate,-,original,0.5,text_labels,standard_palette,-,-,-,-,,,
+58,0,0,0,semantic_segmentation,-,separate,-,original,0.5,text_labels,random,-,-,-,-,,,
+59,0,0,0,semantic_segmentation,-,separate,-,original,1,none,standard_palette,-,-,-,-,,,
+60,0,0,0,semantic_segmentation,-,separate,-,original,1,none,random,-,-,-,-,,,
+61,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,standard_palette,-,-,-,-,,,
+62,0,0,0,semantic_segmentation,-,separate,-,original,1,text_labels,random,-,-,-,-,,,
+63,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+64,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+65,0,0,1,referring_segmentation,-,overlay,-,sub_sampled,-,-,-,-,-,-,-,,,
+66,0,0,1,referring_segmentation,-,separate,-,sub_sampled,-,-,-,-,-,-,-,,,
+67,0,0,1,referring_segmentation,-,text,-,sub_sampled,-,-,-,-,-,-,-,,,
+68,0,1,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill,-,-,,,
+69,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,contour,-,-,,,
+70,0,0,1,referring_segmentation,-,overlay,-,original,0.5,-,-,-,fill_contour,-,-,,,
+71,0,0,1,referring_segmentation,-,overlay,-,original,1,-,-,-,fill,-,-,,,
+72,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,contour,-,-,,,
+73,0,0,0,referring_segmentation,-,overlay,-,original,1,-,-,-,fill_contour,-,-,,,
+74,0,0,1,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill,-,-,,,
+75,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,contour,-,-,,,
+76,0,0,0,referring_segmentation,-,separate,-,original,0.5,-,-,-,fill_contour,-,-,,,
+77,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill,-,-,,,
+78,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,contour,-,-,,,
+79,0,0,0,referring_segmentation,-,separate,-,original,1,-,-,-,fill_contour,-,-,,,
+80,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,polygon,-,,,
+81,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,rle,-,,,
+82,0,0,0,keypoint,points_only,overlay,-,-,-,-,same_color,-,-,-,-,,,
+83,0,0,1,keypoint,points_only,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+84,0,0,0,keypoint,points_only,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+85,0,0,0,keypoint,points_only,separate,-,-,-,-,same_color,-,-,-,-,,,
+86,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+87,0,0,0,keypoint,points_only,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+88,0,0,1,keypoint,skeleton,overlay,-,-,-,-,same_color,-,-,-,-,,,
+89,0,1,1,keypoint,skeleton,overlay,-,-,-,-,color_by_instance,-,-,-,-,,,
+90,0,0,1,keypoint,skeleton,overlay,-,-,-,-,color_by_part,-,-,-,-,,,
+91,0,0,0,keypoint,skeleton,separate,-,-,-,-,same_color,-,-,-,-,,,
+92,0,0,1,keypoint,skeleton,separate,-,-,-,-,color_by_instance,-,-,-,-,,,
+93,0,0,0,keypoint,skeleton,separate,-,-,-,-,color_by_part,-,-,-,-,,,
+94,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,flat_list,-,,,
+95,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,part_keyed_json,-,,,
+96,0,0,1,keypoint,-,-,-,-,-,-,-,-,-,coco_style,-,,,
+97,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,plasma,,,
+98,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,turbo,,,
+99,0,0,1,depth_estimation,-,-,-,-,-,-,-,-,-,-,gray,,,
+100,0,1,1,low_level,-,-,-,-,-,-,-,-,-,-,-,,,
+101,0,1,1,generation,-,-,-,-,-,-,-,-,-,-,-,,,
+1742,0,0,1,instance_segmentation,-,-,-,-,-,-,-,-,-,-,-,42,17,1742
+305,0,0,1,object_detection,-,-,-,-,-,-,-,-,-,-,-,5,3,0305
+7080,0,0,1,referring_segmentation,-,-,-,-,-,-,-,-,-,-,-,80,70,7080
+4649,1,0,1,semantic_segmentation,-,-,-,-,-,-,-,-,-,-,-,46,49,4649

docs/my_drop_tabs/drop_bottom_10pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,350,132,0.6238
+generation_controllable,336,40,0.5642
+generation_editing,302,40,0.6342
+generation_inpainting_high_level,294,50,0.6495
+generation_inpainting_low_level,375,80,0.5288
+generation_t2i,342,342,0.6661
+instance_segmentation,341,35,0.5822
+keypoint,365,87,0.5512
+lowlevel-deblur,258,40,0.673
+lowlevel-derain,391,40,0.7534
+lowlevel-desnow,386,60,0.7498
+lowlevel-super-resolution,353,40,0.5536
+object_detection,387,39,0.6311
+referring_segmentation,389,40,0.5585

docs/my_drop_tabs/drop_bottom_15pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,343,132,0.6307
+generation_controllable,305,40,0.5946
+generation_editing,247,40,0.7118
+generation_inpainting_high_level,246,50,0.7231
+generation_inpainting_low_level,375,80,0.5288
+generation_t2i,342,342,0.6661
+instance_segmentation,337,35,0.5859
+keypoint,348,87,0.5641
+lowlevel-deblur,254,40,0.6791
+lowlevel-derain,369,40,0.782
+lowlevel-desnow,382,60,0.7547
+lowlevel-super-resolution,288,40,0.6179
+object_detection,383,39,0.6348
+referring_segmentation,380,40,0.5653

docs/my_drop_tabs/drop_bottom_20pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,322,130,0.6501
+generation_controllable,268,39,0.6307
+generation_editing,244,40,0.7165
+generation_inpainting_high_level,237,50,0.7395
+generation_inpainting_low_level,306,79,0.5836
+generation_t2i,293,293,0.7296
+instance_segmentation,317,35,0.6025
+keypoint,327,85,0.579
+lowlevel-deblur,254,40,0.6791
+lowlevel-derain,368,40,0.7832
+lowlevel-desnow,382,60,0.7547
+lowlevel-super-resolution,288,40,0.6179
+object_detection,367,39,0.6483
+referring_segmentation,355,39,0.5821

docs/my_drop_tabs/drop_bottom_25pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,296,125,0.6742
+generation_controllable,232,39,0.6668
+generation_editing,241,40,0.7204
+generation_inpainting_high_level,232,50,0.7474
+generation_inpainting_low_level,306,79,0.5836
+generation_t2i,292,292,0.7307
+instance_segmentation,281,35,0.6298
+keypoint,293,84,0.6026
+lowlevel-deblur,240,40,0.6968
+lowlevel-derain,353,40,0.8005
+lowlevel-desnow,375,60,0.7618
+lowlevel-super-resolution,250,39,0.6549
+object_detection,347,39,0.6633
+referring_segmentation,320,39,0.6036

docs/my_drop_tabs/drop_bottom_30pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,273,120,0.6949
+generation_controllable,219,39,0.681
+generation_editing,205,40,0.7717
+generation_inpainting_high_level,214,50,0.7742
+generation_inpainting_low_level,303,79,0.5851
+generation_t2i,292,292,0.7307
+instance_segmentation,271,34,0.6374
+keypoint,252,82,0.631
+lowlevel-deblur,206,40,0.7411
+lowlevel-derain,335,40,0.8205
+lowlevel-desnow,350,60,0.7856
+lowlevel-super-resolution,223,39,0.6823
+object_detection,338,39,0.6696
+referring_segmentation,306,39,0.6117

docs/my_drop_tabs/drop_bottom_35pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,273,120,0.6949
+generation_controllable,219,39,0.681
+generation_editing,205,40,0.7717
+generation_inpainting_high_level,214,50,0.7742
+generation_inpainting_low_level,203,77,0.6622
+generation_t2i,254,254,0.7759
+instance_segmentation,229,31,0.6698
+keypoint,252,82,0.631
+lowlevel-deblur,206,40,0.7411
+lowlevel-derain,335,40,0.8205
+lowlevel-desnow,350,60,0.7856
+lowlevel-super-resolution,223,39,0.6823
+object_detection,300,38,0.6958
+referring_segmentation,254,39,0.6419

docs/my_drop_tabs/drop_bottom_40pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,210,115,0.7534
+generation_controllable,219,39,0.681
+generation_editing,205,40,0.7717
+generation_inpainting_high_level,214,50,0.7742
+generation_inpainting_low_level,203,77,0.6622
+generation_t2i,254,254,0.7759
+instance_segmentation,210,30,0.6852
+keypoint,191,76,0.6728
+lowlevel-deblur,169,40,0.7939
+lowlevel-derain,318,40,0.8377
+lowlevel-desnow,332,60,0.801
+lowlevel-super-resolution,200,39,0.7032
+object_detection,287,37,0.7047
+referring_segmentation,234,39,0.6541

docs/my_drop_tabs/drop_bottom_5pct.csv

@@ -0,0 +1,15 @@
+task,count_questions,count_unique_image_id,accuracy
+depth_estimation,365,133,0.6069
+generation_controllable,352,40,0.5468
+generation_editing,325,40,0.5999
+generation_inpainting_high_level,380,50,0.535
+generation_inpainting_low_level,400,80,0.5046
+generation_t2i,382,382,0.6115
+instance_segmentation,360,38,0.5625
+keypoint,389,87,0.5315
+lowlevel-deblur,258,40,0.673
+lowlevel-derain,391,40,0.7534
+lowlevel-desnow,386,60,0.7498
+lowlevel-super-resolution,353,40,0.5536
+object_detection,400,39,0.6174
+referring_segmentation,399,40,0.5501

docs/paper/doc.md

@@ -0,0 +1,343 @@
+# Judge Dataset Documentation
+
+## Overview
+
+The Judge dataset evaluates how well vision-language models (VLMs) can act as judges of computer vision model outputs. Each prompt presents a VLM with one or more encoded predictions and asks it to assess quality — via pairwise comparison, ranking, or absolute scoring. The dataset covers 15 vision tasks, 53 encoding variants, and three question types.
+
+**Dataset location:** `Icey444/Judge_questions_v2` on Hugging Face  
+**v2.1 split:** 5,549 items (2,855 pairwise · 2,647 scoring · 47 ranking)
+
+---
+
+## 1. Task Coverage
+
+| Task | Category | What is judged |
+|---|---|---|
+| `object_detection` | Perception | Bounding boxes (label + coordinates) for specified classes |
+| `instance_segmentation` | Perception | Per-instance pixel masks for specified classes |
+| `semantic_segmentation` | Perception | Per-class pixel masks across all categories |
+| `referring_segmentation` | Perception | Mask of the region referred to by a natural-language expression |
+| `keypoint` | Perception | 17-keypoint COCO pose skeleton per person |
+| `depth_estimation` | Perception | Dense monocular depth map |
+| `lowlevel-deblur` | Restoration | Image deblurring result |
+| `lowlevel-derain` | Restoration | Image deraining result |
+| `lowlevel-desnow` | Restoration | Image desnowing result |
+| `lowlevel-super-resolution` | Restoration | Image super-resolution result |
+| `generation_controllable` | Generation | Controllable image generation (control condition + prompt) |
+| `generation_editing` | Generation | Instruction-based image editing |
+| `generation_inpainting_high_level` | Generation | High-level inpainting (semantic fill) |
+| `generation_inpainting_low_level` | Generation | Low-level inpainting (seamless texture fill) |
+| `generation_t2i` | Generation | Text-to-image generation |
+
+---
+
+## 2. Encoding Variants
+
+Encoding variants determine **how a model's prediction is presented** to the judge. Each variant transforms raw annotation output (bounding boxes, masks, keypoints, etc.) into a form the VLM can read. There are two broad families: **pixel** (rendered image) and **text** (structured string). Combo encodings pair one text and one pixel form per option.
+
+### 2.1 Object Detection (6 variants)
+
+| Stem | Type | Description |
+|---|---|---|
+| `pixel_s0_m0` | pixel | Colored bounding boxes, no label text, overlaid on original image |
+| `pixel_s1_m0` | pixel | Colored boxes + label text, overlaid on original image |
+| `pixel_s1_m1` | pixel | Colored boxes + label text, rendered on separate black canvas |
+| `0305` | combo | `text_xyxy` coordinates + `pixel_s1_m0` box image, per option |
+| `text_xyxy` | text | `{"label":…,"bbox":[x1,y1,x2,y2]}` per detection |
+| `text_xywh` | text | `{"label":…,"bbox":[x,y,w,h]}` per detection |
+
+### 2.2 Instance Segmentation (12 variants)
+
+**Sub-sampled pixel** (downsampled grid, each cell = most-covered instance index):
+
+| Stem | Embed | Description |
+|---|---|---|
+| `pixel_ss0_m0` | overlay | Grid overlaid on original image |
+| `pixel_ss0_m1` | separate | Grid on black canvas |
+| `pixel_ss1_m0_o0_l0_c0_b0` | overlay | … (sub-sampled text) |
+
+**Original-resolution pixel** (full-resolution mask rendering):
+
+| Stem | Opacity | Label overlay | Color scheme | Bbox style |
+|---|---|---|---|---|
+| `pixel_ss1_m0_o0_l1_c0_b1` | 0.5 | text labels | color-by-class | dashed bbox |
+| `pixel_ss1_m0_o0_l1_c1_b1` | 0.5 | text labels | color-by-instance | dashed bbox |
+| `pixel_ss1_m0_o1_l1_c0_b1` | 1.0 | text labels | color-by-class | dashed bbox |
+| `pixel_ss1_m1_o0_l1_c0_b1` | 0.5 | text labels | color-by-class | dashed bbox, separate canvas |
+| `pixel_ss1_m0_o0_l1_c0_b0` | 0.5 | text labels | color-by-class | no bbox |
+| `pixel_ss1_m0_o0_l0_c0_b1` | 0.5 | no labels | color-by-class | dashed bbox |
+
+**Text-only:**
+
+| Stem | Format |
+|---|---|
+| `text_polygon` | `{"instance_id":…,"label":…,"polygon":[[x,y],…]}` per instance |
+| `text_rle` | `{"instance_id":…,"label":…,"rle":{…}}` COCO RLE per instance |
+| `text_matrix` | 2D integer grid (rows × cols), each cell = instance index |
+
+**Combo:** `1742` — polygon text + color-by-instance image, per option.
+
+### 2.3 Semantic Segmentation (8 variants)
+
+**Sub-sampled pixel** (3): overlay, separate canvas, text sub-sample.
+
+**Original-resolution pixel** (4, all opacity 0.5):
+
+| Stem | Label overlay | Color scheme | Canvas |
+|---|---|---|---|
+| `pixel_ss1_m0_o0_l1_c0` | text labels | standard palette | overlay |
+| `pixel_ss1_m0_o0_l1_c1` | text labels | random colors | overlay |
+| `pixel_ss1_m0_o1_l1_c0` | text labels | standard palette | overlay, full opacity |
+| `pixel_ss1_m1_o0_l1_c0` | text labels | standard palette | separate canvas |
+| `pixel_ss1_m0_o0_l0_c0` | no labels | standard palette | overlay |
+
+**Text-only:**
+
+| Stem | Format |
+|---|---|
+| `text_polygon` | `{"label":…,"polygon":[[x,y],…]}` per segment |
+| `text_matrix` | 2D integer grid, each cell = class index |
+
+**Combo:** `4649` — sub-sample text + original-res overlay image, per option.
+
+### 2.4 Referring Segmentation (11 variants)
+
+**Sub-sampled pixel** (3): overlay, separate, text sub-sample.
+
+**Original-resolution pixel** (5):
+
+| Stem | Mask style | Opacity | Canvas |
+|---|---|---|---|
+| `pixel_ss1_m0_o0_m0` | filled region | 0.5 | overlay |
+| `pixel_ss1_m0_o0_m1` | contour only | 0.5 | overlay |
+| `pixel_ss1_m0_o0_m2` | fill + contour | 0.5 | overlay |
+| `pixel_ss1_m0_o1_m0` | filled region | 1.0 | overlay |
+| `pixel_ss1_m1_o0_m0` | filled region | 0.5 | separate canvas |
+
+**Text-only:**
+
+| Stem | Format |
+|---|---|
+| `text_polygon` | `{"label":"<expression>","polygon":[[x,y],…]}` |
+| `text_matrix` | 2D grid; legend maps index → referring expression |
+
+**Combo:** `7080` — polygon text + fill+contour image, per option.
+
+### 2.5 Keypoint Detection (8 variants)
+
+**Pixel:**
+
+| Stem | Style | Color scheme | Canvas |
+|---|---|---|---|
+| `pixel_s0_c1_m0` | points only | color-by-instance | overlay |
+| `pixel_s1_c0_m0` | skeleton | single color (green) | overlay |
+| `pixel_s1_c1_m0` | skeleton | color-by-instance | overlay |
+| `pixel_s1_c2_m0` | skeleton | color-by-body-part | overlay |
+| `pixel_s1_c1_m1` | skeleton | color-by-instance | separate canvas |
+
+**Text-only:**
+
+| Stem | Format |
+|---|---|
+| `text_flat_list` | 34 numbers `[x0..x16, y0..y16]` per person (COCO order) |
+| `text_part_keyed_json` | `{"person_id":…,"keypoints":[{"name":…,"x":…,"y":…},…]}` |
+| `text_coco_style` | 51 numbers `[x,y,v]×17` per person |
+
+All text formats include the note: *x=0.0, y=0.0 means the keypoint was not detected or is not visible.*
+
+### 2.6 Depth Estimation (3 variants)
+
+Each variant is a colormap applied to the predicted depth map:
+
+| Stem | Colormap | Semantics |
+|---|---|---|
+| `plasma` | Magma/plasma | Bright yellow = closest, dark purple = farthest |
+| `turbo` | Turbo (rainbow) | Red = closest, blue = farthest |
+| `gray` | Grayscale | Bright = closest, dark = farthest |
+
+### 2.7 Low-level Restoration (4 tasks × 1 variant each)
+
+Each task has a single pixel encoding: the restored output image shown alongside the degraded input.
+
+| Task | Input context |
+|---|---|
+| `lowlevel-deblur` | Blurry source image |
+| `lowlevel-derain` | Rainy source image |
+| `lowlevel-desnow` | Snowy source image |
+| `lowlevel-super-resolution` | Low-resolution source image |
+
+### 2.8 Image Generation (5 tasks × 1 variant each)
+
+Each task shows the generated output image(s) alongside the source context.
+
+| Task | Source context shown |
+|---|---|
+| `generation_controllable` | Source image (control signal + reference) |
+| `generation_editing` | Original image before editing |
+| `generation_inpainting_high_level` | Original image with masked region |
+| `generation_inpainting_low_level` | Original image with masked region |
+| `generation_t2i` | No source image (text prompt only) |
+
+---
+
+## 3. Question Types
+
+### 3.1 Pairwise Comparison
+
+The judge sees two options (A and B) and selects the better prediction.
+
+**Structure:**
+```
+[<image>]                        ← original/reference image (if available)
+
+You are a judge to decide the quality of answers to a <task> task [based on my given image].
+[Task-specific context: class(es) of interest / referring prompt / etc.]
+
+Format of predictions: <encoding description>
+
+Options:
+
+A. [<image>] [text or legend]
+
+B. [<image>] [text or legend]
+
+<Final question>. Please answer with A or B.
+```
+
+**Pair sampling:** Within each `(image_id, class-of-interest, error_type, prompt)` group, pairs are drawn so that no two annotations with equal `final_score` are paired. Up to 10 pairs per group per task (encoding_analysis) or 50 pairs (judge_analysis).
+
+**Answer:** The letter corresponding to the annotation with the higher `final_score`.
+
+**Final question phrasing** is sampled from five paraphrases to reduce positional bias:
+- *Which prediction is better?*
+- *Which option is a better execution of the vision task?*
+- *Which option would you prefer as answer to the vision task?*
+- *Which of the two is the better result?*
+- *Which option better fulfills the task?*
+
+### 3.2 Ranking
+
+The judge sees N options (A through E, or fewer) and ranks them best-to-worst.
+
+**Structure:**
+```
+[original image context]
+
+You are a judge to decide the quality of answers to a <task> task.
+[Task-specific context]
+
+Format of predictions: <encoding description>
+
+Options:
+
+A. [<image> or text]
+B. [<image> or text]
+...
+
+Rank the predictions from best to worst. Respond with the ranking as a single string
+of letters only (best first, worst last). For example, BCAED.
+```
+
+Groups of 3–5 annotations sharing the same `(image_id, class-of-interest, error_type)` are ranked together. Used in `judge_analysis` only.
+
+**Answer:** Letters ordered by descending `final_score`.
+
+### 3.3 Scoring
+
+The judge sees a single prediction and assigns a score from 0 to 10.
+
+**Structure:**
+```
+[original image]
+
+You are a judge to decide the quality of answers to a <task> task [based on my given image].
+[Task-specific context]
+
+Format of prediction: <encoding description>
+
+[First image: original. Second image: encoded prediction.]   ← pixel encodings
+[Prediction (text): <content>]                               ← text encodings
+
+Score the quality of the prediction from 0 to 10.
+0 = random guessing / worst, 10 = best possible.
+Please answer with a single score from 0 to 10 only.
+```
+
+**Answer:** The annotation's `final_score` (normalized to 0–10).
+
+Used in `judge_analysis` only. 20 groups × 5 annotations per group × stems per task.
+
+---
+
+## 4. Prompt Construction Standards
+
+### 4.1 Role Framing
+
+Every prompt begins with a judge role sentence tailored to the task:
+
+| Task group | Intro pattern |
+|---|---|
+| Object detection | "You are a judge to decide the quality of answers to an object detection task. **The class(es) of interest is** {coi}." |
+| Instance / semantic segmentation | Same pattern with respective task name and COI |
+| Referring segmentation | "… **The prompt is** '{expression}'." (or "The prompt is the referring expression shown in the options below." if not available at prompt-level) |
+| Keypoint | "… **The task is pose estimation.**" |
+| Depth estimation | "… **The task is depth prediction.**" |
+| Low-level restoration | Task-specific sentence describing the restoration goal |
+| Generation | Task-specific sentence describing the generation goal + text prompt when available |
+
+"based on my given image" is appended when the original image is included as a `<image>` placeholder.
+
+### 4.2 Format Description
+
+After the role sentence, the prompt includes a **Format of predictions** block describing the encoding so the judge knows what it is looking at:
+
+- **Pixel encodings:** describe the visual style (overlay/canvas, color scheme, opacity, label style).
+- **Text encodings:** describe the schema (e.g., JSON structure, coordinate conventions, grid dimensions and legend).
+- **Combo encodings:** each option shows its own format description inline, followed by the encoded image.
+- **Generation/low-level:** no format description (the prediction is a natural image); the instruction covers the task criterion instead.
+
+### 4.3 Color Legends
+
+For encodings where colors carry semantic meaning, a legend is included **per option** (not once globally), because different predictions may contain different classes or instances:
+
+- **Object detection pixel:** legend lists each detected class and its assigned color.
+- **Instance segmentation (color-by-class):** legend lists each class and color.
+- **Instance segmentation (color-by-instance):** no per-instance legend (color only distinguishes people; skeleton structure is self-evident).
+- **Semantic segmentation:** legend lists each class and color.
+- **Keypoint (color-by-part):** legend lists each of the 17 COCO keypoint names and its color.
+- **Keypoint (color-by-instance):** one sentence describing that all keypoints and links of the same person share a color; no per-person list.
+- **Keypoint (same color):** "All keypoints and links use a single color (green). No color legend."
+- **Depth colormaps:** the colormap semantics (which end is near/far) are described in the format block.
+
+### 4.4 Image Placeholder Ordering
+
+`<image>` placeholders in the prompt correspond to `media` entries in the same order:
+
+1. **Original/reference image** (first, when present) — always `original_{image_id}.png`.
+2. **Option A image** (prediction rendered for annotation A).
+3. **Option B image** (prediction rendered for annotation B).
+
+Text-only encodings include only the original image (1 image total). Generation/low-level pixel encodings that have no source image in the image index include 2 images (A and B only). Generation tasks with a retrievable source image include 3 images.
+
+### 4.5 Subset Labels
+
+Each item carries a `subset` field indicating which run produced it:
+
+| Subset | Stems used | Samples | Pairs | Scoring/Ranking |
+|---|---|---|---|---|
+| `encoding_analysis` | All `is_base_ablation=1` (51 regular + gen/lowlevel) | 10 per task | 10 per task | No |
+| `judge_analysis` | `is_final=1` (53) | 20 per task | 50 per task | Yes (20 groups × 5 annotations) |
+
+Both runs use `seed=42`. The v2.1 JSON is the deduplicated union (keyed on task + encoding + question_type + annotation IDs).
+
+---
+
+## 5. Data Sources
+
+| Task group | Annotation source | Image source |
+|---|---|---|
+| Perception tasks | `data_json_v2/*_annotations.json` | `images_v2/*.json` → local files or HF URLs |
+| Low-level restoration | `data_json_v2/lowlevel_annotations.json` | HF: `Icey444/VisualJudge_images` (prediction URLs) + `images_v2/lowlevel.json` (source URLs) |
+| Generation | `data_json_v2/generation_annotations.json` | HF: `Icey444/VisualJudge_images` (prediction URLs) + `images_v2/generation_images.json` (source URLs) |
+
+Predictions are encoded locally by task-specific encoder scripts (`src/encoders/encode_*.py`) and stored under `output/encoded_v2/`. Original images are cached as `original_{image_id}.png` in the same directory.

docs/paper/figs/fig_encoding_top3.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59a43c6cccea2ebcbfd401ea219033eab483d82809d0691b438593bc13016830
+size 1759309