Update README.md

README.md

@@ -160,33 +160,37 @@ dataloader = DataLoader(
 
 # Iterate over the full dataset and store features each time 16278 input images have been processed
 slide_features = []
+
 for i, (slide_ids, tile_ids, imgs) in tqdm(
-      enumerate(dataloader),
-      total=(num_slides*num_tiles)//batch_size+1,
-      desc=f"Extracting features"
-    ):
-    # Set the reference slide
-    reference_slide_id = slide_ids[0]
-
-    # Check whether all slide ids are the same inside the batch.
-    # If yes, it means we have not reached the last batch for a given slide_id.
-    if all(slide_id == reference_slide_id for slide_id in slide_ids):
+    enumerate(dataloader),
+    total=ceil(num_slides * num_tiles / batch_size),
+    desc="Extracting features"
+):
+    # Detect the current slide
+    current_slide_id = slide_ids[0]
+
+    # If all tiles from the current batch belongs to the same slide,
+    # we just add the batch features to the running list
+    if all(slide_id == current_slide_id for slide_id in slide_ids):
         batch_stack = process_imgs(imgs, tile_ids)
         slide_features.append(batch_stack)
-    # Otherwise, a batch contains features of 2 different slides.
-    # First, we need to save the features from slide N-1, and then
-    # initialize the running `slide_features` list with the features from slide N.
+        # If `slide_features` contains 16,278 tiles features, we export it
+        if len(slide_features) == ceil(num_tiles / batch_size):
+            save_features(slide_features, slide_id=current_slide_id)
+            # And initialize `slide_features` as an empty list
+            slide_features = []
+            continue
+    # The current batch contains tiles from current (N) and N+1 slide
     else:
-        # Save features matrix for `reference_slide_id`
-        mask = (np.array(slide_ids) != reference_slide_id)
+        # We get the maximum index at which tiles all belong to slide N
+        mask = (np.array(slide_ids) != current_slide_id)
         idx = mask.argmax()
+        # Then export the slide features with the remaining idx features
         batch_stack = process_imgs(imgs[:idx], tile_ids[:idx])
         slide_features.append(batch_stack)
-        save_features(slide_features, slide_id=reference_slide_id)
-
-        # Initialize `slide_features` with the next slide
-        batch_stack = process_imgs(imgs[idx:], tile_ids[idx:])
-        slide_features = [batch_stack]
+        save_features(slide_features, slide_id=current_slide_id)
+        # And initialize with (batch_size - idx) tiles features from slide N+1
+        slide_features = [process_imgs(imgs[idx:], tile_ids[idx:])]
 ```
 
 # License