Fix inference issue with large image (#368)

mmrotate/apis/inference.py

@@ -49,19 +49,20 @@ def inference_detector_by_patches(model,
     sizes, steps = get_multiscale_patch(sizes, steps, ratios)
     windows = slide_window(width, height, sizes, steps)
 
-    # prepare patch data
-    patch_datas = []
-    for window in windows:
-        data = dict(img=img, win=window.tolist())
-        # build the data pipeline
-        data = test_pipeline(data)
-        patch_datas.append(data)
-
     results = []
     start = 0
     while True:
-        data = patch_datas[start:start + bs]
-        data = collate(data, samples_per_gpu=len(data))
+        # prepare patch data
+        patch_datas = []
+        if (start + bs) > len(windows):
+            end = len(windows)
+        else:
+            end = start + bs
+        for window in windows[start:end]:
+            data = dict(img=img, win=window.tolist())
+            data = test_pipeline(data)
+            patch_datas.append(data)
+        data = collate(patch_datas, samples_per_gpu=len(patch_datas))
         # just get the actual data from DataContainer
         data['img_metas'] = [
             img_metas.data[0] for img_metas in data['img_metas']
@@ -80,7 +81,7 @@ def inference_detector_by_patches(model,
         with torch.no_grad():
             results.extend(model(return_loss=False, rescale=True, **data))
 
-        if start + bs >= len(patch_datas):
+        if end >= len(windows):
             break
         start += bs