key added

app.py

@@ -19,6 +19,7 @@ from langchain.prompts import PromptTemplate
 from torchvision.models import resnet50
 import nest_asyncio
 from sentence_transformers import SentenceTransformer
+from huggingface_hub import hf_hub_download
 
 # model = SentenceTransformer("Alibaba-NLP/gte-Qwen2-1.5B-instruct", trust_remote_code=True)
 
@@ -118,7 +119,11 @@ rag_chain = RetrievalQA.from_chain_type(
 Load My models
 '''
 
-def load_model(model_path):
+def load_model(repo_id, filename):
+    model_path = hf_hub_download(
+        repo_id=repo_id,
+        filename=filename,
+    )
     model = EvoViTModel(img_size=224, patch_size=16, in_channels=3, embed_dim=768, num_classes=2, hidden_dim=512)
     model.classifier = nn.Linear(512, 1)
     state_dict = torch.load(model_path, map_location=device)
@@ -168,9 +173,11 @@ def load_binary_models():
         "Vasculitis Photos": 'keerthi/Vasculitis/best_global_model_10fold.pth',
         "Warts Molluscum and other Viral Infections": 'santhosh/10fold_model_warts.pth'
     }
-    for class_name, rel_path in class_models_mapping.items():
-        model_path = os.path.join("best_models_overall", rel_path)
-        model = load_model(model_path)
+    repo_id = "KeerthiVM/SkinCancerDiagnosis"  # Your Hugging Face repo
+
+    for class_name, filename in class_models_mapping.items():
+        # model_path = os.path.join("best_models_overall", rel_path)
+        model = load_model(repo_id, filename)
         base_models.append(model)
     return base_models
 
@@ -255,8 +262,15 @@ class SkinDiseaseClassifier:
         self.resnet_feature_extractor.eval()
 
         # Load meta model
-        meta_model_path = 'best_meta_model_two_layer_version4.pth'
+        # meta_model_path = 'best_meta_model_two_layer_version4.pth'
+        # checkpoint = torch.load(meta_model_path, map_location=self.device)
+
+        meta_model_path = hf_hub_download(
+            repo_id="KeerthiVM/SkinCancerDiagnosis",
+            filename="best_meta_model_two_layer_version4.pth"
+        )
         checkpoint = torch.load(meta_model_path, map_location=self.device)
+
         correct_input_size = checkpoint['state_dict']['fc.0.weight'].shape[1]
         input_size = 23 + 4 + 7168  # Adjust based on your actual feature size
         fc_layers = [1024, 512, 256]  # Use whatever was in your best model

requirements.txt

@@ -15,4 +15,5 @@ scikit-learn
 langchain_openai
 nest_asyncio
 sentence_transformers
-langchain-qdrant
+langchain-qdrant
+huggingface_hub