Update app.py

app.py

@@ -70,7 +70,7 @@ class resnet_feature_extractor(torch.nn.Module):
 
 
 
-image = Image.open(r'/content/carpet/test/color/000.png')
+image = Image.open(r'carpet/test/color/000.png')
 image = transform(image).unsqueeze(0)
 
 backbone = resnet_feature_extractor()
@@ -85,7 +85,7 @@ print(feature.shape)
 
 memory_bank =[]
 
-folder_path = Path(r'/content/carpet/train/good')
+folder_path = Path(r'carpet/train/good')
 
 for pth in tqdm(folder_path.iterdir(),leave=False):
     with torch.no_grad():
@@ -97,7 +97,7 @@ memory_bank = torch.cat(memory_bank,dim=0)
 
 y_score=[]
 
-folder_path = Path(r'/content/carpet/train/good')
+folder_path = Path(r'carpet/train/good')
 
 for pth in tqdm(folder_path.iterdir(),leave=False):
     data = transform(Image.open(pth)).unsqueeze(0)
@@ -123,7 +123,7 @@ y_score = []
 y_true=[]
 
 for classes in ['color','good','cut','hole','metal_contamination','thread']:
-    folder_path = Path(r'/content/carpet/test/{}'.format(classes))
+    folder_path = Path(r'carpet/test/{}'.format(classes))
 
     for pth in tqdm(folder_path.iterdir(),leave=False):
 
@@ -150,7 +150,7 @@ plt.vlines(x=best_threshold,ymin=0,ymax=30,color='r')
 plt.show()
 
 
-test_image = transform(Image.open(r'/content/carpet/test/color/000.png')).unsqueeze(0)
+test_image = transform(Image.open(r'carpet/test/color/000.png')).unsqueeze(0)
 features = backbone(test_image)
 
 distances = torch.cdist(features, memory_bank, p=2.0)