Update app.py

app.py

@@ -1,40 +1,32 @@
 import gradio as gr
-import torch
-import numpy
+import cv2
+import numpy as np
 from PIL import Image
 
 rng = range(1,225)
-sobel_verticle = torch.tensor([[-1,0,1],
+sobel_verticle = np.array([[-1,0,1],
                                [-1,0,1],
-                               [-1,0,1]]).float()
+                               [-1,0,1]])
                             
-sobel_horizontal = torch.tensor([[-1,-1,11],
+sobel_horizontal = np.array([[-1,-1,11],
                                  [ 0, 0, 0],
-                                 [ 1, 1, 1]]).float()
+                                 [ 1, 1, 1]])
 
-laplacian = torch.tensor([[ 0,-1, 0],
+laplacian = np.array([[ 0,-1, 0],
                           [-1, 5,-1],
-                          [ 0,-1, 0]]).float()
+                          [ 0,-1, 0]])
 
-average = torch.tensor([[1/9,1/9, 1/9],
+average = np.array([[1/9,1/9, 1/9],
                           [1/9,1/9, 1/9],
-                          [1/9,1/9, 1/9]]).float()
+                          [1/9,1/9, 1/9]])
 
 
-weighted_average = torch.tensor([[1/16,2/16, 1/16],
+weighted_average = np.array([[1/16,2/16, 1/16],
                                  [2/16,4/16, 2/16],
-                                 [1/16,2/16, 1/16]]).float()
-
-
-def apply_kernel(img, row, col, kernel):
-    return (img[row-1:row+2,col-1:col+2] * kernel).sum()
+                                 [1/16,2/16, 1/16]])
 
 def convolve(img, kernel):
-    ch1, ch2, ch3 = img[:,:,0], img[:,:,1], img[:,:,2] 
-    ch1 = torch.tensor([[apply_kernel(ch1, i,j,kernel) for j in rng] for i in rng])
-    ch2 = torch.tensor([[apply_kernel(ch2, i,j,kernel) for j in rng] for i in rng])
-    ch3 = torch.tensor([[apply_kernel(ch3, i,j,kernel) for j in rng] for i in rng])
-    out_img = (torch.stack((ch1,ch2,ch3),-1)).numpy()
+    out_img = cv2.filter2D(src=img, kernel=kernel)
     return Image.fromarray(out_img)