def __next__()

in pytorchvideo/data/charades.py [0:0]

• 39 lines of code
• 10 McCabe index (conditional complexity)

    def __next__(self) -> dict:
        """
        Retrieves the next clip based on the clip sampling strategy and video sampler.

        Returns:
            A dictionary with the following format.

            .. code-block:: text

                {
                    'video': <video_tensor>,
                    'label': <index_label>,
                    'video_label': <index_label>
                    'video_index': <video_index>,
                    'clip_index': <clip_index>,
                    'aug_index': <aug_index>,
                }
        """
        if not self._video_sampler_iter:
            # Setup MultiProcessSampler here - after PyTorch DataLoader workers are spawned.
            self._video_sampler_iter = iter(MultiProcessSampler(self._video_sampler))

        if self._loaded_video:
            video, video_index = self._loaded_video
        else:
            video_index = next(self._video_sampler_iter)
            path_to_video_frames = self._path_to_videos[video_index]
            video = FrameVideo.from_frame_paths(path_to_video_frames)
            self._loaded_video = (video, video_index)

        clip_start, clip_end, clip_index, aug_index, is_last_clip = self._clip_sampler(
            self._next_clip_start_time, video.duration, {}
        )
        # Only load the clip once and reuse previously stored clip if there are multiple
        # views for augmentations to perform on the same clip.
        if aug_index == 0:
            self._loaded_clip = video.get_clip(clip_start, clip_end, self._frame_filter)

        frames, frame_indices = (
            self._loaded_clip["video"],
            self._loaded_clip["frame_indices"],
        )
        self._next_clip_start_time = clip_end

        if is_last_clip:
            self._loaded_video = None
            self._next_clip_start_time = 0.0

        # Merge unique labels from each frame into clip label.
        labels_by_frame = [
            self._labels[video_index][i]
            for i in range(min(frame_indices), max(frame_indices) + 1)
        ]
        sample_dict = {
            "video": frames,
            "label": labels_by_frame,
            "video_label": self._video_labels[video_index],
            "video_name": str(video_index),
            "video_index": video_index,
            "clip_index": clip_index,
            "aug_index": aug_index,
        }
        if self._transform is not None:
            sample_dict = self._transform(sample_dict)

        return sample_dict