def __getitem__()

in dataset/dataset_downstream.py [0:0]

• 27 lines of code
• 8 McCabe index (conditional complexity)

    def __getitem__(self, item):
        frame_root, frame_num, cls = self.samples[item]
        sample_num = 3 if self.sample_num <= 0 or self.sample_num > frame_num else self.sample_num

        frame_indices = np.round(np.linspace(1, frame_num, num=frame_num)).astype(np.int64)
        if self.data_source == 'anet':
            frame_indices = np.round(np.linspace(0, frame_num-1, num=frame_num)).astype(np.int64)

        segments_length = frame_num // sample_num
        segments = []

        for i in range(sample_num):
            start_idx = i * segments_length
            if i == sample_num - 1:
                segment = frame_indices[start_idx:]
            else:
                end = (i + 1) * segments_length
                segment = frame_indices[start_idx:end]

            segments.append(segment)

        images = []
        images_ids = []

        for segment in segments:
            image_path_ind = np.random.choice(segment, 1)[0]
            image = self._get_aug_frame(frame_root, image_path_ind).unsqueeze(dim=0)
            images.append(image)
            images_ids.append(image_path_ind)

        if len(images) < sample_num:
            return None

        clips = torch.cat(images, dim=0)

        return clips, cls