mindspore.dataset.vision.read_video

mindspore.dataset.vision.read_video(filename, start_pts=0, end_pts=None, pts_unit='pts')

Read the video, audio, metadata from a video file.

It supports AVI, H264, H265, MOV, MP4, WMV file formats on CPU, and H264, H265 file formats on Ascend.

Note

This method is executed on CPU by default, but it is also supported to be executed on Ascend by setting video backend with mindspore.dataset.config.set_video_backend("Ascend") .

Parameters

• filename (str) – The path to the video file to be read.

• start_pts (Union[float, Fraction, int], optional) – The start presentation timestamp of the video. Default: 0, read from the beginning.

• end_pts (Union[float, Fraction, int], optional) – The end presentation timestamp of the video. Default: None, read until the end.

• pts_unit (str, optional) – The unit of the timestamps. It can be any of ["pts", "sec"]. Default: "pts".

Returns

• numpy.ndarray, four dimensions uint8 data for video. The format is [T, H, W, C]. T is the number of frames, H is the height, W is the width, C is the channel for RGB.

• numpy.ndarray, two dimensions float for audio. The format is [C, L]. C is the number of channels. L is the length of the points in one channel.

• dict, metadata for the video and audio. It contains video_fps data of type float and audio_fps data of type int.

Raises

• TypeError – If filename is not of type str.

• TypeError – If start_pts is not of type [float, Fraction, int].

• TypeError – If end_pts is not of type [float, Fraction, int].

• TypeError – If pts_unit is not of type str.

• RuntimeError – If filename does not exist, or not a regular file, or not a supported video file.

• ValueError – If start_pts is less than 0.

• ValueError – If end_pts is less than start_pts.

• ValueError – If pts_unit is not in ["pts", "sec"].

Supported Platforms:

CPU Ascend

Examples

>>> import mindspore.dataset.vision as vision
>>> video_output, audio_output, metadata_output = vision.read_video("/path/to/file")