correction bug affichage gradcam online

app/model.py

@@ -20,7 +20,7 @@ import tensorflow as tf
 from tf_keras_vis.saliency import Saliency
 from tf_keras_vis.utils import normalize
 import logging
-
+import time
 
 from typing import TypedDict, Callable, Any
 logging.basicConfig(
@@ -352,17 +352,31 @@ def predict_with_model(config, image_bytes: bytes,show_heatmap=False):
         "entropy":entropy,
         "is_uncertain_model":is_uncertain_model
     }
-    
     if show_heatmap and not is_uncertain_model:
-        # Génère la heatmap uniquement si demandé
-        #heatmap = make_gradcam_heatmap(config["model"],input_array)
+        try:
+            logger.info("✅ Début de la génération de la heatmap")
+            start_time = time.time()
 
-        heatmap = compute_gradcam(config["gradcam_model"],raw_input,class_index=predicted_class_index,layer_name=config["last_conv_layer"])
-       
-        result["heatmap"] = heatmap.tolist()
-    else:
-        result["heatmap"]=[]
+            # Vérification des entrées
+            logger.info(f"🖼️ Image d'entrée shape: {raw_input.shape}")
+            logger.info(f"🎯 Index de classe prédite: {predicted_class_index}")
+            logger.info(f"🛠️ Dernière couche utilisée: {config['last_conv_layer']}")
+
+            # Calcul de la heatmap
+            heatmap = compute_gradcam(config["gradcam_model"], raw_input, class_index=predicted_class_index, layer_name=config["last_conv_layer"])
 
+            elapsed_time = time.time() - start_time
+            logger.info(f"✅ Heatmap générée en {elapsed_time:.2f} secondes")
+
+            # Conversion en liste pour le JSON
+            result["heatmap"] = heatmap.tolist()
+
+        except Exception as e:
+            logger.error(f"❌ Erreur lors de la génération de la heatmap: {e}")
+            result["heatmap"] = []
+    else:
+        logger.info("ℹ️ Heatmap non générée (option désactivée ou modèle incertain)")
+        result["heatmap"] = []
 
 
     return result

app/voting.py

@@ -75,11 +75,11 @@ async def soft_voting(model_configs,image_bytes: bytes,mode,show_heatmap):
 
             if show_heatmap:
                 heatmap = prediction.get("heatmap")
-                if isinstance(heatmap, np.ndarray) and heatmap.size > 0:
+                if heatmap and len(heatmap) > 0:
                     models_heatmaps.append(heatmap)
                 else:
                     logger.warning(f"⚠️ Heatmap vide ou invalide, non ajoutée pour le modèle {config['model_name']}")
-                #logger.info(f"Taille heatmaps :{len(models_heatmaps)}")    
+                logger.info(f"Taille heatmaps :{len(models_heatmaps)}")    
             
             models.append(config["model_name"])
             logger.info(f"📊 Prédictions ajoutées pour {config['model_name']}")