Skip to main content

Ray Data video API

ray.data.read_datasource​

ray.data.read_datasource(
    datasource: VideoDatasource,
    *,
    paths: str
) -> Dataset
info

This API depends on decord. To install it, run pip install decord.

Read videos into a Ray Dataset. Ray Data stores each frame in a separate row.

This Datasource supports any video format supported by Decord. Supported formats include MP4, MOV, and AVI. For the list of supported formats, see the FFMPEG documentation.

Parameters​

  • datasource: A VideoDatasource.
  • paths: A file path or list of file paths to read video files from.

Returns​

A Ray Dataset that contains the video frames.

Examples​

import ray
from ray.anyscale.data import VideoDatasource

ds = ray.data.read_datasource(
    VideoDatasource(),
    paths="basketball.mp4"
)