Update utils.py

Add normalisation before passing image to model during test

utils.py

@@ -524,13 +524,25 @@ def plot_couple_examples(model, loader, thresh, iou_thresh, anchors):
         plot_image(x[i].permute(1,2,0).detach().cpu(), nms_boxes)
         
 def plot_single_image(model, image, thresh, iou_thresh, anchors):
-    print('calling to plot single image')
     model.eval()
-    with torch.no_grad():
-        x = image
-        print('Train begin')
+    with torch.no_grad():      
+              
+        import albumentations as A
+        from albumentations.pytorch import ToTensorV2
+        import config
+        #print('----------------')
+        #print('type of input image before transform = ',type(image))
+        test_transforms = A.Compose(
+            [                          
+                A.Normalize(mean=[0, 0, 0], std=[1, 1, 1], max_pixel_value=255,),
+                ToTensorV2(),
+            ])       
+        
+        img = test_transforms(image=image)
+        print('type of input image after transform = ',type(img))
+        x = img['image'].unsqueeze(0)
         out = model(x)
-        print('Train end')
+        print('Train end, out size = ',out[0].size())
         bboxes = [[] for _ in range(x.shape[0])]
         for i in range(3):
             batch_size, A, S, _, _ = out[i].shape
@@ -542,12 +554,13 @@ def plot_single_image(model, image, thresh, iou_thresh, anchors):
                 bboxes[idx] += box
 
         model.train()
-    print('batch size = ',batch_size)
+   
     for i in range(batch_size):
+        print('total bboxes input to nms = ', len(bboxes[i][0]))        
         nms_boxes = non_max_suppression(
             bboxes[i], iou_threshold=iou_thresh, threshold=thresh, box_format="midpoint",
         )
-        print('calling to plot bounding boxes image')
+        print('calling to plot bounding boxes image, nms_boxes = ',nms_boxes)
         fig = plot_image(x[i].permute(1,2,0).detach().cpu(), nms_boxes) 
         return fig