clip_len (int): Frames of each sampled output clip.含义

鱼儿会飞吗

于 2024-08-24 14:41:49 发布

阅读量150

点赞数 8

文章标签：深度学习 pytorch 人工智能 python

本文链接：https://blog.csdn.net/qq_34425255/article/details/141499433

版权

class UniformSampleFrames:
    """Uniformly sample frames from the video.

    To sample an n-frame clip from the video. UniformSampleFrames basically
    divide the video into n segments of equal length and randomly sample one
    frame from each segment. To make the testing results reproducible, a
    random seed is set during testing, to make the sampling results
    deterministic.

    Required keys are "total_frames", "start_index" , added or modified keys
    are "frame_inds", "clip_len", "frame_interval" and "num_clips".

    Args:
        clip_len (int): Frames of each sampled output clip.
        num_clips (int): Number of clips to be sampled. Default: 1.
        seed (int): The random seed used during test time. Default: 255.
    """

在"UniformSampleFrames"这个采样方法中,clip_len这个参数就代表了要从视频中采样出的每个子clip的帧数n。

也就是说,clip_len参数指定了:

将整个视频时长均匀地划分成多少个等长的时间段
在每个时间段内随机采样出一个帧

这样就可以得到clip_len个采样帧,组成一个子clip。重复这个过程,就可以从整个视频中采样出多个这样的子clip。