[Improvement] Use Pylint to polish code style (open-mmlab#908)

* polish

* polish

* polish tools

* polish tests

* polish demo

* remove topk class

.pylintrc

@@ -60,7 +60,9 @@ confidence=
 # --enable=similarities". If you want to run only the classes checker, but have
 # no Warning level messages displayed, use "--disable=all --enable=classes
 # --disable=W".
-disable=print-statement,
+disable=import-outside-toplevel
+        redefined-outer-name
+        print-statement,
         parameter-unpacking,
         unpacking-in-except,
         old-raise-syntax,

demo/demo_gradcam.py

@@ -78,7 +78,7 @@ def build_inputs(model, video_path, use_frames=False):
     if osp.isfile(video_path) and use_frames:
         raise RuntimeError(
             f"'{video_path}' is a video file, not a rawframe directory")
-    elif osp.isdir(video_path) and not use_frames:
+    if osp.isdir(video_path) and not use_frames:
         raise RuntimeError(
             f"'{video_path}' is a rawframe directory, not a video file")
 

demo/long_video_demo.py

@@ -215,7 +215,7 @@ def inference(model, data, args, frame_queue):
 
     if args.stride > 0:
         pred_stride = int(args.sample_length * args.stride)
-        for i in range(pred_stride):
+        for _ in range(pred_stride):
             frame_queue.popleft()
 
     # for case ``args.stride=0``

demo/webcam_demo.py

@@ -77,7 +77,7 @@ def show_results():
     cur_time = time.time()
     while True:
         msg = 'Waiting for action ...'
-        ret, frame = camera.read()
+        _, frame = camera.read()
         frame_queue.append(np.array(frame[:, :, ::-1]))
 
         if len(result_queue) != 0:
@@ -93,7 +93,7 @@ def show_results():
                 cv2.putText(frame, text, location, FONTFACE, FONTSCALE,
                             FONTCOLOR, THICKNESS, LINETYPE)
 
-        elif len(text_info):
+        elif len(text_info) != 0:
             for location, text in text_info.items():
                 cv2.putText(frame, text, location, FONTFACE, FONTSCALE,
                             FONTCOLOR, THICKNESS, LINETYPE)

demo/webcam_demo_spatiotemporal_det.py

@@ -223,7 +223,6 @@ def _do_detect(self, image):
 
         The format of bboxes is (xmin, ymin, xmax, ymax) in pixels.
         """
-        pass
 
     def predict(self, task):
         """Add keyframe bboxes to task."""
@@ -571,20 +570,20 @@ def __next__(self):
         if self.read_queue.qsize() == 0:
             time.sleep(0.02)
             return not self.stopped, None
-        else:
-            was_read, task = self.read_queue.get()
-            if not was_read:
-                # If we reach the end of the video, there aren't enough frames
-                # in the task.processed_frames, so no need to model inference
-                # and draw predictions. Put task into display queue.
-                with self.read_id_lock:
-                    read_id = self.read_id
-                with self.display_lock:
-                    self.display_queue[read_id] = was_read, copy.deepcopy(task)
-
-                # main thread doesn't need to handle this task again
-                task = None
-            return was_read, task
+
+        was_read, task = self.read_queue.get()
+        if not was_read:
+            # If we reach the end of the video, there aren't enough frames
+            # in the task.processed_frames, so no need to model inference
+            # and draw predictions. Put task into display queue.
+            with self.read_id_lock:
+                read_id = self.read_id
+            with self.display_lock:
+                self.display_queue[read_id] = was_read, copy.deepcopy(task)
+
+            # main thread doesn't need to handle this task again
+            task = None
+        return was_read, task
 
     def start(self):
         """Start read thread and display thread."""
@@ -685,9 +684,9 @@ def draw_clip_range(self, frames, preds, bboxes, draw_range):
     @abstractmethod
     def draw_one_image(self, frame, bboxes, preds):
         """Draw bboxes and corresponding texts on one frame."""
-        pass
 
-    def abbrev(self, name):
+    @staticmethod
+    def abbrev(name):
         """Get the abbreviation of label name:
 
         'take (an object) from (a person)' -> 'take ... from ...'

mmaction/apis/train.py

@@ -225,5 +225,5 @@ def train_model(model,
 
                 eval_res = test_dataset.evaluate(outputs, **eval_cfg)
                 runner.logger.info(f'Testing results of the {name} checkpoint')
-                for name, val in eval_res.items():
-                    runner.logger.info(f'{name}: {val:.04f}')
+                for metric_name, val in eval_res.items():
+                    runner.logger.info(f'{metric_name}: {val:.04f}')

mmaction/core/bbox/transforms.py

@@ -15,22 +15,22 @@ def bbox2result(bboxes, labels, num_classes, thr=0.01):
     """
     if bboxes.shape[0] == 0:
         return list(np.zeros((num_classes - 1, 0, 5), dtype=np.float32))
-    else:
-        bboxes = bboxes.cpu().numpy()
-        labels = labels.cpu().numpy()
 
-        # We only handle multilabel now
-        assert labels.shape[-1] > 1
+    bboxes = bboxes.cpu().numpy()
+    labels = labels.cpu().numpy()
 
-        scores = labels  # rename for clarification
-        thr = (thr, ) * num_classes if isinstance(thr, float) else thr
-        assert scores.shape[1] == num_classes
-        assert len(thr) == num_classes
+    # We only handle multilabel now
+    assert labels.shape[-1] > 1
 
-        result = []
-        for i in range(num_classes - 1):
-            where = scores[:, i + 1] > thr[i + 1]
-            result.append(
-                np.concatenate((bboxes[where, :4], scores[where, i + 1:i + 2]),
-                               axis=1))
-        return result
+    scores = labels  # rename for clarification
+    thr = (thr, ) * num_classes if isinstance(thr, float) else thr
+    assert scores.shape[1] == num_classes
+    assert len(thr) == num_classes
+
+    result = []
+    for i in range(num_classes - 1):
+        where = scores[:, i + 1] > thr[i + 1]
+        result.append(
+            np.concatenate((bboxes[where, :4], scores[where, i + 1:i + 2]),
+                           axis=1))
+    return result

mmaction/core/evaluation/ava_evaluation/np_box_list.py

@@ -120,8 +120,9 @@ def get_coordinates(self):
         x_max = box_coordinates[:, 3]
         return [y_min, x_min, y_max, x_max]
 
-    def _is_valid_boxes(self, data):
-        """Check whether data fullfills the format of N*[ymin, xmin, ymax,
+    @staticmethod
+    def _is_valid_boxes(data):
+        """Check whether data fulfills the format of N*[ymin, xmin, ymax,
         xmin].
 
         Args:
@@ -131,7 +132,7 @@ def _is_valid_boxes(self, data):
             a boolean indicating whether all ymax of boxes are equal or greater
             than ymin, and all xmax of boxes are equal or greater than xmin.
         """
-        if len(data):
+        if len(data) != 0:
             for v in data:
                 if v[0] > v[2] or v[1] > v[3]:
                     return False

mmaction/core/evaluation/ava_evaluation/object_detection_evaluation.py

@@ -29,6 +29,7 @@
 
 import collections
 import logging
+import warnings
 from abc import ABCMeta, abstractmethod
 from collections import defaultdict
 
@@ -101,15 +102,13 @@ def clear(self):
 class ObjectDetectionEvaluator(DetectionEvaluator):
     """A class to evaluate detections."""
 
-    def __init__(
-        self,
-        categories,
-        matching_iou_threshold=0.5,
-        evaluate_corlocs=False,
-        metric_prefix=None,
-        use_weighted_mean_ap=False,
-        evaluate_masks=False,
-    ):
+    def __init__(self,
+                 categories,
+                 matching_iou_threshold=0.5,
+                 evaluate_corlocs=False,
+                 metric_prefix=None,
+                 use_weighted_mean_ap=False,
+                 evaluate_masks=False):
         """Constructor.
 
         Args:
@@ -244,7 +243,8 @@ def add_single_detected_image_info(self, image_id, detections_dict):
             detected_masks=detection_masks,
         )
 
-    def create_category_index(self, categories):
+    @staticmethod
+    def create_category_index(categories):
         """Creates dictionary of COCO compatible categories keyed by category
         id.
 
@@ -277,14 +277,8 @@ def evaluate(self):
             2. per_category_ap: category specific results with keys of the form
                'PerformanceByCategory/mAP@<matching_iou_threshold>IOU/category'
         """
-        (
-            per_class_ap,
-            mean_ap,
-            _,
-            _,
-            per_class_corloc,
-            mean_corloc,
-        ) = self._evaluation.evaluate()
+        (per_class_ap, mean_ap, _, _, per_class_corloc,
+         mean_corloc) = self._evaluation.evaluate()
 
         metric = f'mAP@{self._matching_iou_threshold}IOU'
         pascal_metrics = {self._metric_prefix + metric: mean_ap}
@@ -355,15 +349,13 @@ def __init__(self, categories, matching_iou_threshold=0.5):
 class ObjectDetectionEvaluation:
     """Internal implementation of Pascal object detection metrics."""
 
-    def __init__(
-        self,
-        num_groundtruth_classes,
-        matching_iou_threshold=0.5,
-        nms_iou_threshold=1.0,
-        nms_max_output_boxes=10000,
-        use_weighted_mean_ap=False,
-        label_id_offset=0,
-    ):
+    def __init__(self,
+                 num_groundtruth_classes,
+                 matching_iou_threshold=0.5,
+                 nms_iou_threshold=1.0,
+                 nms_max_output_boxes=10000,
+                 use_weighted_mean_ap=False,
+                 label_id_offset=0):
         if num_groundtruth_classes < 1:
             raise ValueError(
                 'Need at least 1 groundtruth class for evaluation.')
@@ -399,13 +391,11 @@ def _initialize_detections(self):
     def clear_detections(self):
         self._initialize_detections()
 
-    def add_single_ground_truth_image_info(
-        self,
-        image_key,
-        groundtruth_boxes,
-        groundtruth_class_labels,
-        groundtruth_masks=None,
-    ):
+    def add_single_ground_truth_image_info(self,
+                                           image_key,
+                                           groundtruth_boxes,
+                                           groundtruth_class_labels,
+                                           groundtruth_masks=None):
         """Adds groundtruth for a single image to be used for evaluation.
 
         Args:
@@ -420,8 +410,8 @@ def add_single_ground_truth_image_info(
                 masks. The mask values range from 0 to 1.
         """
         if image_key in self.groundtruth_boxes:
-            logging.warn(('image %s has already been added to the ground '
-                          'truth database.'), image_key)
+            warnings.warn(('image %s has already been added to the ground '
+                           'truth database.'), image_key)
             return
 
         self.groundtruth_boxes[image_key] = groundtruth_boxes
@@ -430,14 +420,12 @@ def add_single_ground_truth_image_info(
 
         self._update_ground_truth_statistics(groundtruth_class_labels)
 
-    def add_single_detected_image_info(
-        self,
-        image_key,
-        detected_boxes,
-        detected_scores,
-        detected_class_labels,
-        detected_masks=None,
-    ):
+    def add_single_detected_image_info(self,
+                                       image_key,
+                                       detected_boxes,
+                                       detected_scores,
+                                       detected_class_labels,
+                                       detected_masks=None):
         """Adds detections for a single image to be used for evaluation.
 
         Args:
@@ -468,8 +456,8 @@ def add_single_detected_image_info(
             )
 
         if image_key in self.detection_keys:
-            logging.warn(('image %s has already been added to the ground '
-                          'truth database.'), image_key)
+            warnings.warn(('image %s has already been added to the ground '
+                           'truth database.'), image_key)
             return
 
         self.detection_keys.add(image_key)
@@ -536,8 +524,7 @@ def evaluate(self):
             logging.info(
                 'The following classes have no ground truth examples: %s',
                 np.squeeze(np.argwhere(self.num_gt_instances_per_class == 0)) +
-                self.label_id_offset,
-            )
+                self.label_id_offset)
 
         if self.use_weighted_mean_ap:
             all_scores = np.array([], dtype=float)
@@ -557,10 +544,8 @@ def evaluate(self):
                 all_scores = np.append(all_scores, scores)
                 all_tp_fp_labels = np.append(all_tp_fp_labels, tp_fp_labels)
             precision, recall = metrics.compute_precision_recall(
-                scores,
-                tp_fp_labels,
-                self.num_gt_instances_per_class[class_index],
-            )
+                scores, tp_fp_labels,
+                self.num_gt_instances_per_class[class_index])
             self.precisions_per_class.append(precision)
             self.recalls_per_class.append(recall)
             average_precision = metrics.compute_average_precision(
@@ -569,8 +554,7 @@ def evaluate(self):
 
         self.corloc_per_class = metrics.compute_cor_loc(
             self.num_gt_imgs_per_class,
-            self.num_images_correctly_detected_per_class,
-        )
+            self.num_images_correctly_detected_per_class)
 
         if self.use_weighted_mean_ap:
             num_gt_instances = np.sum(self.num_gt_instances_per_class)