fix issues lambda

main.py

@@ -16,6 +16,24 @@ from typing import List
 import httpx
 from concurrent.futures import ProcessPoolExecutor
 
+
+def process_single_image(image_url, model):
+    try:
+        response = requests.get(image_url)
+        image = Image.open(BytesIO(response.content))
+        processed_image = process_image(image, size=image_size)
+        image_tensor = transforms.ToTensor()(processed_image).unsqueeze(0)
+
+        with torch.no_grad():
+            outputs = model(image_tensor)
+            probabilities = torch.nn.functional.softmax(outputs, dim=1)
+            predicted_probabilities = probabilities.numpy().tolist()
+            confidence = round(predicted_probabilities[0][1], 2)
+        return {"imageUrl": image_url, "confidence": confidence}
+    except Exception as e:
+        return {"imageUrl": image_url, "error": str(e)}
+
+
 app = FastAPI()
 
 image_size = 256
@@ -61,23 +79,6 @@ model_pipelines = {}
 base_model = ResNet("resnet152", num_output_neurons=2).to(device)
 
 
-def process_single_image(image_url, model):
-    try:
-        response = requests.get(image_url)
-        image = Image.open(BytesIO(response.content))
-        processed_image = process_image(image, size=image_size)
-        image_tensor = transforms.ToTensor()(processed_image).unsqueeze(0)
-
-        with torch.no_grad():
-            outputs = model(image_tensor)
-            probabilities = torch.nn.functional.softmax(outputs, dim=1)
-            predicted_probabilities = probabilities.numpy().tolist()
-            confidence = round(predicted_probabilities[0][1], 2)
-        return {"imageUrl": image_url, "confidence": confidence}
-    except Exception as e:
-        return {"imageUrl": image_url, "error": str(e)}
-
-
 @app.on_event("startup")
 async def load_models():
     # Charger les modèles au démarrage
@@ -194,7 +195,7 @@ async def batch_predict(request: BatchPredictRequest):
     with ProcessPoolExecutor() as executor:
         results = list(
             executor.map(
-                lambda url: process_single_image(url, model), request.imageUrls
+                process_single_image, [(url, model) for url in request.imageUrls]
             )
         )