livesync-io 0.3.1

The graph-based video processing framework

LiveSync

A Keras-inspired asynchronous stream processing framework for building real-time media applications. LiveSync provides a flexible layer system for creating both synchronous and asynchronous media processing pipelines.

Installation

We recommend using rye for installation:

rye add livesync-io

Alternatively, you can use pip:

pip install livesync-io

Quick Start

Here's a simple example of a webcam recording pipeline:

import livesync as ls
from livesync import layers

x = ls.WebcamInput(device_id=0, fps=30)

f1 = layers.FpsControlLayer(fps=10)
f2 = layers.VideoRecorderLayer(filename="./examples/output.mp4", fps=f1.fps)

h = f1(x)
y = f2(h)

sync = ls.Sync(inputs=[x], outputs=[y])
with sync.compile() as runner:
    runner.run(callback=ls.LoggingCallback())

You can visualize the pipeline by printing y.graph:

>>> print(y.graph)

● (f2): Records processed frames to MP4 file
│
● (f1): Controls frame rate of the stream
│
◇ (x): Captures frames from webcam (device 0)

Features

• Layer-Based Architecture: Build complex processing pipelines using a Keras-inspired layer system
• Async-First Design: Built from the ground up for asynchronous stream processing
• Media Processing: Optimized for real-time audio and video processing
• Flexible Stream System: Support for both synchronous and asynchronous data flows
• Remote Processing: Built-in gRPC support for distributed processing

Requirements

• Python 3.10 or higher
• OpenCV Python
• FFmpeg
• gRPC tools (for remote processing)

Documentation

For detailed documentation and examples, visit our documentation site.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

LiveSync is developed and maintained by OS Designers, Inc. For support, feature requests, or bug reports, please open an issue on our GitHub repository.