pupil-labs-video 1.0.6

A high-level wrapper of PyAV providing an easy to use interface to video data.

Pupil Labs Video

A high-level wrapper of PyAV providing an easy to use interface to video data.

The goal of this library is to provide a simple interface while maintaining good computational performance. At current, only MP4 and MJPEG videos are officially compatible.

Features include:

• Performant reading of video files (optionally including audio) utilizing multi-threading.
• Ability to arbitrarily index frames by their index or timestamp.
• Ability to slice the video by frame index or time. Large slices will be loaded lazily to avoidexcessive RAM consumption.
• A frame buffer is maintained to cache frames close to the current decoding position, which avoids repetitive seeking when going back and forth between frames in the same neighborhood or iterating backwards.
• Avoids demuxing and seeking operations as much as possible.
• Reading multi-part video files (e.g. how they are generated by Neon or Pupil Invisible).

Installation

pip install pupil-labs-video

or

pip install -e git+https://github.com/pupil-labs/pl-video.git

Quick Start

You can open a video file and read frames like this:

import pupil_labs.video as plv

with plv.Reader(video_path) as video:
    # Iterate through video frames
    for frame in video:
        # Convert video frame to BGR array
        img = frame.bgr

    # Index individual frames or slices
    first_frame = video[0]
    last_frame = video[-1]
    frames = video[10:20]

    # Index frames by time
    ts = video[10].time
    frame = video.by_container_timestamps[ts]
    frames = video.by_container_timestamps[ts : ts + 10]

You can write video files like this:

import pupil_labs.video as plv

with (plv.Writer(out_path) as writer):
    for img in images:
        writer.write_image(img)