Upload model artifacts and classifier scripts

README.md

@@ -23,12 +23,15 @@
 
         ## Classes
 
-
+        - `Complex Table`
+- `Simple Table`
+- `Visual - Essential`
+- `Visual - Supportive`
 
         Pages matching any of the following classes should trigger image embedding:
 
-        - `Visual Essential`
-- `Complex Table`
+        - `Complex Table`
+- `Visual - Essential`
 
         Default threshold: `0.5`
 

config.json

@@ -1,128 +1,27 @@
 {
-  "model": {
-    "name": "efficientnet_lite0",
-    "pretrained": true,
-    "dropout": 0,
-    "use_spatial_pooling": false
-  },
-  "classes": [
-    "Visual - Essential",
-    "Simple Table",
-    "Chart/Graph",
-    "Visual - Supportive",
-    "Annotated figure",
-    "No Specific Feature",
-    "Diagram/Flowchart",
-    "Visual - Decorative",
+  "image_size": 224,
+  "mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "center_crop_shortest": true,
+  "whiteout_header": false,
+  "whiteout_fraction": 0.15,
+  "class_names": [
     "Complex Table",
-    "Infographic",
-    "Form",
-    "Text to OCR"
+    "Simple Table",
+    "Visual - Essential",
+    "Visual - Supportive"
   ],
-  "class_mapping": {
-    "Form": null,
-    "No Specific Feature": null,
-    "Text to OCR": null,
-    "Visual - Decorative": null,
-    "Infographic": null,
-    "Chart/Graph": null,
-    "Annotated figure": null,
-    "Diagram/Flowchart": null
-  },
+  "threshold": 0.5,
   "image_required_classes": [
-    "Visual Essential",
-    "Complex Table"
-  ],
-  "data": {
-    "train_split": 0.8,
-    "val_split": 0.1,
-    "test_split": 0.1,
-    "image_size": 224,
-    "batch_size": 32,
-    "num_workers": 4,
-    "seed": 42
-  },
-  "augmentation": {
-    "center_crop_shortest": true,
-    "whiteout_header": false,
-    "whiteout_fraction": 0.15,
-    "train": {
-      "horizontal_flip": 0.5,
-      "rotation_degrees": 5,
-      "color_jitter": {
-        "brightness": 0.2,
-        "contrast": 0.2,
-        "saturation": 0.1,
-        "hue": 0.05
-      },
-      "random_erasing": 0.1
-    },
-    "val": {
-      "enabled": false
-    }
-  },
-  "training": {
-    "epochs": 40,
-    "learning_rate": 0.0001,
-    "weight_decay": 0.0001,
-    "optimizer": "adamw",
-    "scheduler": "cosine",
-    "warmup_epochs": 5,
-    "label_smoothing": 0.0,
-    "gradient_clip_norm": 1.0,
-    "pos_weight": [
-      3.6715595722198486,
-      6.668674468994141,
-      2.3281044960021973,
-      6.0722222328186035
-    ]
-  },
-  "monitoring": {
-    "metric": "val_f1",
-    "mode": "max"
-  },
-  "early_stopping": {
-    "enabled": true,
-    "patience": 20
-  },
-  "evaluation": {
-    "threshold": 0.5,
-    "save_confusion_matrix": true,
-    "save_per_class_metrics": true
-  },
-  "checkpointing": {
-    "save_best_only": true,
-    "save_last": true
-  },
-  "paths": {
-    "data_dir": "data",
-    "output_dir": "outputs",
-    "checkpoint_dir": "checkpoints",
-    "logs_dir": "logs"
-  },
-  "logging": {
-    "use_tensorboard": false,
-    "use_wandb": true,
-    "wandb_project": "pdf-page-classifier",
-    "log_interval": 10,
-    "wandb_run_name": "silver-line-69"
-  },
-  "qat": {
-    "enabled": true,
-    "epochs": 5,
-    "learning_rate": "1e-5",
-    "preset": "mixed",
-    "num_init_samples": 300
-  },
-  "onnx": {
-    "opset_version": 14,
-    "dynamic_axes": true,
-    "simplify": true,
-    "input_names": [
-      "image"
-    ],
-    "output_names": [
-      "probabilities"
-    ]
-  }
+    "Complex Table",
+    "Visual - Essential"
+  ]
 }

model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:049ad13d69105d029d2b03c26863f6c9f6338ae303679967a51e7a4d6af21271
+size 13461047

model_int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6cfd45f2c34606f96166056019a991509267f754d7b37785ac0fa974cb0b0be
+size 3812747

openvino_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32751981326b62f992d5d5de7db6a05bd9bf8def264bde1ba6bab8b61568b1d4
+size 6710280

openvino_model.xml

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:595d8436a54945623bb3f21f9457b253d07a0bef3b5de4255183cb6ba03c3c68
+size 197760

openvino_model_int8.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433c7fcc0fece5d3f14df1ce8a83f3856e642bb3904567a65c88c3048b629a12
+size 3507576

openvino_model_int8.xml

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ca09dfb6c67f71def7da80fc655bd07b20df2dac98e1c4459e5a3624c38f8f6
+size 387908