Upload vit_base_patch16_224 multi-label model (AUC: 0.7595)

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+auroc_comparison_chexnet_vit.png filter=lfs diff=lfs merge=lfs -text
+comparison_chexnet_vs_vit.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,171 @@
+---
+tags:
+- pytorch
+- image-classification
+- medical-imaging
+- chest-x-ray
+- multi-label-classification
+- chexnet
+- vit-base-patch16-224
+license: mit
+datasets:
+- alkzar90/NIH-Chest-X-ray-dataset
+language:
+- en
+metrics:
+- accuracy
+- f1
+- roc_auc
+---
+
+# 🫁 ViT-Base-Patch16-224 - Multi-Label Chest X-ray Classification (14 Pathologies)
+
+Ce modèle a été entraîné pour la classification multi-label de **14 pathologies thoraciques** 
+à partir de radiographies X-ray du dataset ChestX-ray14.
+
+## 📋 Description
+
+- **Architecture**: ViT-Base-Patch16-224
+- **Tâche**: Classification multi-label (14 pathologies)
+- **Dataset**: [NIH Chest X-ray (ChestX-ray14)](https://huggingface.co/datasets/alkzar90/NIH-Chest-X-ray-dataset)
+- **Framework**: PyTorch
+- **Image Size**: 224×224
+
+## 📊 Performance Globale
+
+| Métrique | Valeur |
+|----------|--------|
+| **AUC-ROC (macro)** | **0.7595** |
+| AUC-ROC (micro) | 0.8191 |
+| F1 (macro) | 0.0916 |
+| mAP | 0.2041 |
+
+### Comparaison avec l'article CheXNet
+
+| Modèle | AUC Macro | Δ |
+|--------|-----------|---|
+| CheXNet (article) | 0.8414 | - |
+| **Notre modèle** | **0.7595** | -0.0819 |
+
+## 📈 Performance par Pathologie
+
+| Pathologie | AUROC | F1 | Support |
+|------------|-------|----|---------|
+| Atelectasis | 0.7253 | 0.0457 | 3279 |
+| Cardiomegaly | 0.8585 | 0.2737 | 1069 |
+| Effusion | 0.7918 | 0.3613 | 4658 |
+| Infiltration | 0.6662 | 0.2098 | 6112 |
+| Mass | 0.7563 | 0.1563 | 1748 |
+| Nodule | 0.6822 | 0.0322 | 1623 |
+| Pneumonia | 0.6705 | 0.0000 | 555 |
+| Pneumothorax | 0.7889 | 0.0729 | 2665 |
+| Consolidation | 0.7221 | 0.0129 | 1815 |
+| Edema | 0.8229 | 0.0796 | 925 |
+| Emphysema | 0.7516 | 0.0159 | 1093 |
+| Fibrosis | 0.7701 | 0.0000 | 435 |
+| Pleural_Thickening | 0.7429 | 0.0000 | 1143 |
+| Hernia | 0.8836 | 0.0227 | 86 |
+
+
+## 🏷️ Les 14 Pathologies
+
+| ID | Pathologie |
+|----|------------|
+| 0 | Atelectasis |
+| 1 | Cardiomegaly |
+| 2 | Effusion |
+| 3 | Infiltration |
+| 4 | Mass |
+| 5 | Nodule |
+| 6 | Pneumonia |
+| 7 | Pneumothorax |
+| 8 | Consolidation |
+| 9 | Edema |
+| 10 | Emphysema |
+| 11 | Fibrosis |
+| 12 | Pleural_Thickening |
+| 13 | Hernia |
+
+## ⚙️ Configuration d'entraînement
+
+```json
+{
+  "data_variant": "full",
+  "batch_size": 16,
+  "image_size": 224,
+  "num_classes": 14,
+  "learning_rate": 0.0001,
+  "num_epochs": 50,
+  "scheduler": "ReduceLROnPlateau (factor=0.5, patience=5)",
+  "optimizer": "AdamW (weight_decay=0.01)",
+  "loss": "BCEWithLogitsLoss (non pond\u00e9r\u00e9e)"
+}
+```
+
+## 🚀 Utilisation
+
+```python
+import torch
+from torchvision import transforms
+from PIL import Image
+
+# Charger le modèle
+# Pour ViT
+import timm
+
+model = timm.create_model(
+    'vit_base_patch16_224',
+    pretrained=False,
+    num_classes=14
+)
+model.load_state_dict(torch.load('pytorch_model.bin', map_location='cpu'))
+model.eval()
+
+# Préprocessing
+transform = transforms.Compose([
+    transforms.Resize((224, 224)),
+    transforms.ToTensor(),
+    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
+])
+
+# Pathologies
+PATHOLOGIES = [
+    "Atelectasis", "Cardiomegaly", "Effusion", "Infiltration",
+    "Mass", "Nodule", "Pneumonia", "Pneumothorax", "Consolidation",
+    "Edema", "Emphysema", "Fibrosis", "Pleural_Thickening", "Hernia"
+]
+
+# Prédiction
+image = Image.open('chest_xray.png').convert('RGB')
+input_tensor = transform(image).unsqueeze(0)
+
+with torch.no_grad():
+    logits = model(input_tensor)
+    probs = torch.sigmoid(logits)
+
+# Afficher les probabilités
+for name, prob in zip(PATHOLOGIES, probs[0]):
+    print(f"{name}: {prob:.4f}")
+```
+
+## 📚 Citation
+
+```bibtex
+@inproceedings{Wang_2017,
+    title = {ChestX-Ray8: Hospital-Scale Chest X-Ray Database and Benchmarks},
+    author = {Wang, Xiaosong and Peng, Yifan and Lu, Le and Lu, Zhiyong and Bagheri, Mohammadhadi and Summers, Ronald M},
+    booktitle = {IEEE CVPR},
+    year = {2017}
+}
+
+@article{rajpurkar2017chexnet,
+    title={CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning},
+    author={Rajpurkar, Pranav and others},
+    journal={arXiv preprint arXiv:1711.05225},
+    year={2017}
+}
+```
+
+## 📄 License
+
+MIT License

config.json

@@ -0,0 +1,49 @@
+{
+  "model_type": "vit_base_patch16_224",
+  "model_name": "vit_base_patch16_224",
+  "num_classes": 14,
+  "class_names": [
+    "Atelectasis",
+    "Cardiomegaly",
+    "Effusion",
+    "Infiltration",
+    "Mass",
+    "Nodule",
+    "Pneumonia",
+    "Pneumothorax",
+    "Consolidation",
+    "Edema",
+    "Emphysema",
+    "Fibrosis",
+    "Pleural_Thickening",
+    "Hernia"
+  ],
+  "image_size": 224,
+  "task": "multi-label-classification",
+  "dataset": "NIH Chest X-ray (ChestX-ray14)",
+  "training_config": {
+    "data_variant": "full",
+    "batch_size": 16,
+    "image_size": 224,
+    "num_classes": 14,
+    "learning_rate": 0.0001,
+    "num_epochs": 50,
+    "scheduler": "ReduceLROnPlateau (factor=0.5, patience=5)",
+    "optimizer": "AdamW (weight_decay=0.01)",
+    "loss": "BCEWithLogitsLoss (non pond\u00e9r\u00e9e)"
+  },
+  "metrics": {
+    "auc_macro": 0.7594917020663023,
+    "auc_micro": 0.8191216196817769,
+    "auc_weighted": 0.7393044150431767,
+    "f1_macro": 0.09163418365201925,
+    "f1_micro": 0.1665905283995736,
+    "f1_weighted": 0.1486175695741358,
+    "precision_macro": 0.3352479370599724,
+    "precision_micro": 0.48587670989518567,
+    "recall_macro": 0.05992412756010187,
+    "recall_micro": 0.10052929500845402,
+    "mAP": 0.20411386154092143
+  },
+  "uploaded_at": "2025-12-06T23:55:15.974422"
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b88237f468a2998cd08385f2873d1ea59a22b0bcad3dd55fec8ae7bfda29c6d
+size 343300969

run_report.json

@@ -0,0 +1,354 @@
+{
+  "timestamp_utc": "2025-12-06T23:46:39",
+  "model": "ViT (vit_base_patch16_224) Multi-Label",
+  "task": "14 pathologies classification",
+  "config": {
+    "data_variant": "full",
+    "batch_size": 16,
+    "image_size": 224,
+    "num_classes": 14,
+    "learning_rate": 0.0001,
+    "num_epochs": 50,
+    "scheduler": "ReduceLROnPlateau (factor=0.5, patience=5)",
+    "optimizer": "AdamW (weight_decay=0.01)",
+    "loss": "BCEWithLogitsLoss (non pond\u00e9r\u00e9e)"
+  },
+  "training": {
+    "history": {
+      "train_loss": [
+        0.15795161853920808,
+        0.1514771259932308,
+        0.1495829650051838,
+        0.14779357858368056,
+        0.1466357085110594,
+        0.14525542317449325,
+        0.14423545537153284,
+        0.1432988069253473,
+        0.14245807581171868,
+        0.14148768732251146,
+        0.14068052766089545,
+        0.14042458126487148,
+        0.1391056569957907,
+        0.13849249350502918,
+        0.1376091423363124,
+        0.13681426282282455,
+        0.1362750617546518,
+        0.13535684639963436,
+        0.13481243216197838,
+        0.1338201760641047,
+        0.13303599148936676,
+        0.1323197164285935,
+        0.13143239429866188,
+        0.13038955146046127,
+        0.12952067795165,
+        0.12843691301154164,
+        0.1272272804741659,
+        0.12609869775478846,
+        0.12479035782191282,
+        0.12319476159684818,
+        0.121741261945179,
+        0.12002663772751448,
+        0.11790216172227679,
+        0.10852766634013487,
+        0.10346905367254892,
+        0.09948658696596802,
+        0.09563699353869774
+      ],
+      "val_loss": [
+        0.15671193304982953,
+        0.15331958706685223,
+        0.15214708272553637,
+        0.1512915269642132,
+        0.1497921284060932,
+        0.1502516961169331,
+        0.14787093696637207,
+        0.14803033278714486,
+        0.14634910712703758,
+        0.14607930663360244,
+        0.1463124513791361,
+        0.14558507918678018,
+        0.14541163907666965,
+        0.14452245534705807,
+        0.1427664903927312,
+        0.1434842435550117,
+        0.1447336386937421,
+        0.14370530595841116,
+        0.142669265046578,
+        0.1424816575059719,
+        0.14306365069201604,
+        0.14173207662710421,
+        0.14249003686823378,
+        0.14283239976425488,
+        0.1431077379533632,
+        0.14204279520650448,
+        0.14159522234512126,
+        0.14250034649551685,
+        0.14255158479933158,
+        0.1440697303334468,
+        0.14438093068721994,
+        0.1442309853495159,
+        0.1443068087996154,
+        0.14927874723803974,
+        0.15281767301715896,
+        0.1568728258313401,
+        0.1633390645030025
+      ],
+      "train_auc_macro": [
+        0.625451578883831,
+        0.6906600626258499,
+        0.7104018717847923,
+        0.7250215666240262,
+        0.7352947702884968,
+        0.7444551905854568,
+        0.7547180563568351,
+        0.7606128725604394,
+        0.76635652627271,
+        0.7734457640895069,
+        0.7784413391122141,
+        0.7828781532509616,
+        0.7895272460668252,
+        0.7940546568786492,
+        0.7992262888737107,
+        0.8031265032720657,
+        0.8052926494302518,
+        0.810928757860809,
+        0.8142243348282235,
+        0.8207327918213497,
+        0.8241436059952091,
+        0.8264919213437528,
+        0.8319619667793193,
+        0.8359461870519015,
+        0.841056439776067,
+        0.8464174663115808,
+        0.8511028512364415,
+        0.8557113752386379,
+        0.8597649576881208,
+        0.8661855166219318,
+        0.871011514068717,
+        0.8763834117979201,
+        0.8825489035720537,
+        0.9072618315026882,
+        0.9184990290422054,
+        0.9254639864094848,
+        0.9326443951316907
+      ],
+      "val_auc_macro": [
+        0.6919775468440609,
+        0.7316169582279445,
+        0.7315104047459193,
+        0.7398268055941861,
+        0.756912761419654,
+        0.747133598417906,
+        0.7658034727823019,
+        0.7653786783970856,
+        0.7727286506044241,
+        0.7759599770817529,
+        0.7779724208666704,
+        0.7775259925678133,
+        0.7813753581208358,
+        0.7868283739144734,
+        0.7918397356122081,
+        0.7887067843824543,
+        0.7930826156884682,
+        0.7980492438531822,
+        0.7952513916902587,
+        0.7997887894653531,
+        0.8020735566497487,
+        0.8054967472406098,
+        0.806244411412705,
+        0.7958049611152823,
+        0.8056448099858572,
+        0.8053965317067766,
+        0.8103128673428281,
+        0.8003536483912346,
+        0.8006700132190374,
+        0.8057644890223367,
+        0.7998996020776034,
+        0.8044427417405859,
+        0.8032892661538235,
+        0.7935798416729678,
+        0.7928453308685848,
+        0.7859274987555495,
+        0.7827800942977532
+      ],
+      "lr": [
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        0.0001,
+        5e-05,
+        5e-05,
+        5e-05,
+        5e-05,
+        5e-05
+      ]
+    },
+    "best_val_loss": 0.14159522234512126,
+    "best_val_auc": 0.8103128673428281,
+    "best_epoch": 27
+  },
+  "test_metrics": {
+    "global": {
+      "auc_macro": 0.7594917020663023,
+      "auc_micro": 0.8191216196817769,
+      "auc_weighted": 0.7393044150431767,
+      "f1_macro": 0.09163418365201925,
+      "f1_micro": 0.1665905283995736,
+      "f1_weighted": 0.1486175695741358,
+      "precision_macro": 0.3352479370599724,
+      "precision_micro": 0.48587670989518567,
+      "recall_macro": 0.05992412756010187,
+      "recall_micro": 0.10052929500845402,
+      "mAP": 0.20411386154092143
+    },
+    "per_class": {
+      "Atelectasis": {
+        "auc": 0.7253292384102571,
+        "f1": 0.04566473988439306,
+        "precision": 0.43646408839779005,
+        "recall": 0.02409271119243672,
+        "support": 3279
+      },
+      "Cardiomegaly": {
+        "auc": 0.8585349117749351,
+        "f1": 0.2736842105263158,
+        "precision": 0.547752808988764,
+        "recall": 0.1824134705332086,
+        "support": 1069
+      },
+      "Effusion": {
+        "auc": 0.7918250619578521,
+        "f1": 0.3612588015519471,
+        "precision": 0.546284224250326,
+        "recall": 0.2698583082868184,
+        "support": 4658
+      },
+      "Infiltration": {
+        "auc": 0.6662446796525757,
+        "f1": 0.20977622202775348,
+        "precision": 0.4446210916799152,
+        "recall": 0.13727094240837695,
+        "support": 6112
+      },
+      "Mass": {
+        "auc": 0.7562567792049877,
+        "f1": 0.15633672525439407,
+        "precision": 0.4082125603864734,
+        "recall": 0.09668192219679633,
+        "support": 1748
+      },
+      "Nodule": {
+        "auc": 0.6821773771525006,
+        "f1": 0.032238805970149255,
+        "precision": 0.5192307692307693,
+        "recall": 0.0166358595194085,
+        "support": 1623
+      },
+      "Pneumonia": {
+        "auc": 0.6704625315383671,
+        "f1": 0.0,
+        "precision": 0.0,
+        "recall": 0.0,
+        "support": 555
+      },
+      "Pneumothorax": {
+        "auc": 0.7888961934338786,
+        "f1": 0.07288016818500351,
+        "precision": 0.5502645502645502,
+        "recall": 0.03902439024390244,
+        "support": 2665
+      },
+      "Consolidation": {
+        "auc": 0.7220871281481165,
+        "f1": 0.012868632707774798,
+        "precision": 0.24,
+        "recall": 0.006611570247933884,
+        "support": 1815
+      },
+      "Edema": {
+        "auc": 0.8228541662330322,
+        "f1": 0.07955596669750231,
+        "precision": 0.27564102564102566,
+        "recall": 0.046486486486486484,
+        "support": 925
+      },
+      "Emphysema": {
+        "auc": 0.7515767903244963,
+        "f1": 0.01588702559576346,
+        "precision": 0.225,
+        "recall": 0.008234217749313814,
+        "support": 1093
+      },
+      "Fibrosis": {
+        "auc": 0.7700539559718172,
+        "f1": 0.0,
+        "precision": 0.0,
+        "recall": 0.0,
+        "support": 435
+      },
+      "Pleural_Thickening": {
+        "auc": 0.7429438529728626,
+        "f1": 0.0,
+        "precision": 0.0,
+        "recall": 0.0,
+        "support": 1143
+      },
+      "Hernia": {
+        "auc": 0.8836411621525531,
+        "f1": 0.022727272727272728,
+        "precision": 0.5,
+        "recall": 0.011627906976744186,
+        "support": 86
+      }
+    }
+  },
+  "comparison_with_article": {
+    "vit_mean_auc": 0.7594917020663023,
+    "article_mean_auc": 0.8413785714285714,
+    "delta": -0.08188686936226908
+  },
+  "pathology_labels": [
+    "Atelectasis",
+    "Cardiomegaly",
+    "Effusion",
+    "Infiltration",
+    "Mass",
+    "Nodule",
+    "Pneumonia",
+    "Pneumothorax",
+    "Consolidation",
+    "Edema",
+    "Emphysema",
+    "Fibrosis",
+    "Pleural_Thickening",
+    "Hernia"
+  ]
+}