Update app.py

app.py

@@ -7,7 +7,7 @@ import keras.backend as K
 
 from matplotlib import pyplot as plt
 from PIL import Image
-import keras
+from tensorflow import keras
 
 
 resized_shape = (768, 768, 3)
@@ -21,14 +21,13 @@ IMG_SCALING = (1, 1)
 #     gdown.download(url, output, quiet=False)
 #     return output
 
-#model_file = "./seg_unet_model.h5"
-model_file = 'seg_unet_model.h5'
+model_file = "./seg_unet_model.h5"
 
 #Custom objects for model
 
 def Combo_loss(y_true, y_pred, eps=1e-9, smooth=1):
-    targets = keras.ops.cast(K.flatten(y_true), dtype="float32")
-    inputs = keras.ops.cast(K.flatten(y_pred), dtype="float32")
+    targets = tf.dtypes.cast(K.flatten(y_true), tf.float32)
+    inputs = tf.dtypes.cast(K.flatten(y_pred), tf.float32)
     intersection = K.sum(targets * inputs)
     dice = (2. * intersection + smooth) / (K.sum(targets) + K.sum(inputs) + smooth)
     inputs = K.clip(inputs, eps, 1.0 - eps)
@@ -38,20 +37,20 @@ def Combo_loss(y_true, y_pred, eps=1e-9, smooth=1):
     return combo
 
 def dice_coef(y_true, y_pred, smooth=1):
-    y_pred = keras.ops.cast(y_pred, dtype="int32")
-    y_true = keras.ops.cast(y_true, tdtype="int32")
+    y_pred = tf.dtypes.cast(y_pred, tf.int32)
+    y_true = tf.dtypes.cast(y_true, tf.int32)
     intersection = K.sum(y_true * y_pred, axis=[1,2,3])                     
     union = K.sum(y_true, axis=[1,2,3]) + K.sum(y_pred, axis=[1,2,3])           
     return K.mean((2 * intersection + smooth) / (union + smooth), axis=0)
 
 def focal_loss_fixed(y_true, y_pred, gamma=2.0, alpha=0.25):
-    pt_1 = keras.ops.where(keras.ops.equal(y_true, 1), y_pred, keras.ops.ones_like(y_pred))
-    pt_0 = keras.ops.where(keras.ops.equal(y_true, 0), y_pred, keras.ops.ones_like(y_pred))
+    pt_1 = tf.where(tf.equal(y_true, 1), y_pred, tf.ones_like(y_pred))
+    pt_0 = tf.where(tf.equal(y_true, 0), y_pred, tf.zeros_like(y_pred))
     focal_loss_fixed = -K.mean(alpha * K.pow(1. - pt_1, gamma) * K.log(pt_1+K.epsilon())) - K.mean((1 - alpha) * K.pow(pt_0, gamma) * K.log(1. - pt_0 + K.epsilon()))
     return focal_loss_fixed
 
 # Load the model
-seg_model = keras.models.load_model(model_file, custom_objects={'Combo_loss': Combo_loss, 'focal_loss_fixed': focal_loss_fixed, 'dice_coef': dice_coef})
+seg_model = keras.models.load_model('seg_unet_model.h5', custom_objects={'Combo_loss': Combo_loss, 'focal_loss_fixed': focal_loss_fixed, 'dice_coef': dice_coef})
 
 # inputs = gr.inputs.Image(type="pil", label="Upload an image")
 # image_output = gr.outputs.Image(type="pil", label="Output Image")
@@ -68,7 +67,7 @@ def gen_pred(img, model=seg_model):
     img = img[::IMG_SCALING[0], ::IMG_SCALING[1]]
     img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
     img = img/255
-    img = keras.ops.expand_dims(img, axis=0)
+    img = tf.expand_dims(img, axis=0)
     pred = model.predict(img)
     pred = np.squeeze(pred, axis=0)
     fig = plt.figure(figsize=(3, 3))