Update app.py

app.py

@@ -8,6 +8,7 @@ import tensorflow as tf
 import keras
 from keras.models import Model
 from keras.preprocessing import image
+from huggingface_hub import hf_hub_download
 
 # ---- Configuration ----
 CLASS_NAMES = ['akiec', 'bcc', 'bkl', 'df', 'nv', 'vasc', 'mel']
@@ -22,23 +23,15 @@ diagnosis_map = {
     'mel': 'Malin'
 }
 
-from huggingface_hub import hf_hub_download
-import keras
-
-# Télécharge le fichier .keras localement
+# ---- Téléchargement des modèles depuis Hugging Face ----
 resnet_path = hf_hub_download(repo_id="ericjedha/resnet50", filename="Resnet50.keras")
 densenet_path = hf_hub_download(repo_id="ericjedha/densenet201", filename="Densenet201.keras")
 
-# Charge les modèles
+# ---- Chargement modèles ----
 model_resnet50 = keras.saving.load_model(resnet_path, compile=False)
 model_densenet = keras.saving.load_model(densenet_path, compile=False)
-
-
-
-# ---- Chargement modèles ----
 model_xcept    = keras.saving.load_model("Xception.keras", compile=False)
 
-
 # ---- Préprocesseurs ----
 from tensorflow.keras.applications.xception import preprocess_input as preprocess_xception
 from tensorflow.keras.applications.resnet50 import preprocess_input as preprocess_resnet
@@ -54,9 +47,9 @@ def predict_single(img_path, weights=(0.45, 0.25, 0.30)):
     br = preprocess_resnet(np.expand_dims(load_image(img_path, (224, 224)), axis=0))
     bd = preprocess_densenet(np.expand_dims(load_image(img_path, (224, 224)), axis=0))
 
-    pred_x = model_xcept.predict(bx, verbose=0)
-    pred_r = model_resnet50.predict(br, verbose=0)
-    pred_d = model_densenet.predict(bd, verbose=0)
+    pred_x = np.array(model_xcept.predict(bx, verbose=0))
+    pred_r = np.array(model_resnet50.predict(br, verbose=0))
+    pred_d = np.array(model_densenet.predict(bd, verbose=0))
 
     preds = (weights[0] * pred_x + weights[1] * pred_r + weights[2] * pred_d)
 
@@ -75,6 +68,7 @@ def make_gradcam(img_path, model, last_conv_layer_name="conv5_block32_concat", c
 
     if class_index is None:
         preds = model.predict(input_array)
+        preds = np.array(preds)
         class_index = np.argmax(preds[0])
 
     grad_model = Model(inputs=model.inputs, outputs=[
@@ -84,6 +78,12 @@ def make_gradcam(img_path, model, last_conv_layer_name="conv5_block32_concat", c
 
     with tf.GradientTape() as tape:
         conv_outputs, predictions = grad_model(input_array)
+
+        # ✅ Corrige : forcer predictions en Tensor
+        if isinstance(predictions, list):
+            predictions = predictions[0]
+        predictions = tf.convert_to_tensor(predictions)
+
         loss = predictions[:, class_index]
 
     grads = tape.gradient(loss, conv_outputs)[0]