CSTD

Synthetic control route with exactly two texture regions per image, hand-verified.

One of the four evaluation routes in the ICLR 2027 submission on sub-semantic image segmentation: partitioning an image into regions that are coherent in appearance and describable in language, but that need not correspond to any object, part or material class.

• Images: 256
• Code: github.com/aviadcohz/Qwen2SAM_Detecture_Benchmark
• Weights: aviadcohz/Detecture-ICLR-2027
• All four routes in one download: aviadcohz/Detecture_ICLR_Benchmarking

Layout

CSTD/
├── images/            RGB images
├── textures_mask/     per-texture binary masks, <id>_mask_<k>.png
├── metadata.json      image paths, mask paths, descriptions
└── summary.json       dataset statistics

Plus the screening artifacts: verified_256_ids.json, screen_scores.json, screen_rank.json and screen_cstd.py.

The directory inside this repo is named CSTD rather than CSTD, because that is the name the evaluation configs resolve (fairness_baseline_suite/src/paths.py). Paths inside metadata.json are relative to the repository root, so the folder can be placed anywhere.

Use

cd ~/datasets
git lfs install
git clone https://huggingface.co/datasets/aviadcohz/CSTD
mv CSTD/CSTD . && rm -rf CSTD

Then, from the benchmark repo:

cd Qwen2SAM_Detecture_Benchmark/fairness_baseline_suite
PYTHONPATH=src python src/run_fairness.py --model detecture --dataset CSTD

Evaluation protocol

Every number reported on this route comes from one protocol applied identically to every method: no ground-truth region count in the prompt, no inverse-mask completion, no truncation of proposals to a known count, and no dropping of images where a method returns nothing. The region count is inferred, never supplied. Results obtained this way are not comparable to evaluations that supply it.

Read this before using CSTD

This is a 256-image hand-verified subset, not CSTD as published. The original is on Kaggle as architexanonymous/cstd-controlnet-synthetic-texture.

CSTD's released regions/*.png is the stitching mask fed into ControlNet, not an annotation of what came out. Where the generator invented a third material or drifted from the mask, the ground truth silently stops describing the image. Three failure modes were confirmed by eye: one region holding two distinct textures, a third material at an edge or corner, and a contour sitting on no real appearance change.

All 10,000 images were screened on texture features and ranked; 1,296 candidates were reviewed by eye and 274 accepted, a 21% pass rate. The top 256 form this subset. screen_cstd.py and screen_scores.json ship here so the selection is reproducible rather than asserted.

Anyone evaluating on CSTD as published will get different numbers, and should.

Licence

CC-BY-4.0 for this packaging. Upstream corpora keep their own terms: DTD textures and Stable Diffusion 1.5.