Update app .py

app .py

@@ -1,36 +1,43 @@
 
 import gradio as gr
-import tensorflow as tf 
-import tensorflow.keras 
+
 import matplotlib.pyplot as plt
 import cv2
-import tensorflow_io as tfio
+import torch
+import timm
 import numpy as np
 
-loaded_model = tf.keras.models.load_model( 'brain1.h5')
+midas = torch.hub.load('intel-isl/MiDaS', 'MiDaS_small')
+midas.to('cpu')
+midas.eval()
+
+transforms = torch.hub.load('intel-isl/MiDaS', 'transforms')
+transform = transforms.small_transform 
+
+def predict_image(img): 
+  img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+  input_batch = transform(img).to('cpu')
+  with torch.no_grad():
+    prediction = midas(input_batch)
+
+    prediction = torch.nn.functional.interpolate(
+        prediction.unsqueeze(1),
+        size=img.shape[:2],
+        mode="bicubic",
+        align_corners=False,
+    ).squeeze()
 
-def take_img(img):
-  
-  resize = tf.image.resize(img, (128,128))
-  gray = tfio.experimental.color.bgr_to_rgb(resize)
-  yhat = loaded_model.predict(np.expand_dims(gray/255, 0))
-  label_names = {
-  "1": "Tumor",
-  "2": "Normal"}
-  classes_x=np.argmax(yhat,axis=1)
-  a = classes_x[0]
-  input_value = a + 1
-  input_str = str(input_value)
-  predicted_label = label_names[input_str]
-  tumor = yhat[0][0]
-  tumor = str(tumor)
-  normal = yhat[0][1]
-  normal = str(normal)
-  return {'Tumour': tumor, 'Normal':normal}
+  img = prediction.cpu().numpy()
 
+  img = (img / 1000.0)*255
+  out = (img).astype(np.uint8)
+ 
 
+  return out
 
-image = gr.inputs.Image(shape=(128,128))
+    
+image = gr.inputs.Image()
 
 label = gr.outputs.Label('ok')
-gr.Interface(fn=take_img, inputs=image, outputs="label",interpretation='default').launch(debug='True')
+gr.Interface(fn=predict_image, inputs=image, outputs=image).launch(debug='True')