luxonis/depthai-core

What it solves

DepthAI is a library designed to interface with Luxonis hardware, providing a way to build and manage vision pipelines on these devices. It simplifies the process of coordinating multiple hardware devices, managing data streams, and deploying AI nodes to the device's hardware.

How it works

The library allows users to create a Pipeline and add nodes (such as cameras or filters) to it. These nodes can be linked together to form a processing chain. The library supports both C++ and Python bindings, enabling developers to define pipelines that can run either on the device itself or on the host machine. It also supports multi-device configurations where a single pipeline can drive several devices, automatically handling device-to-device links via host relays and managing reconnection logic.

Who it’s for

Developers building applications that require integration with Luxonis hardware for computer vision and AI tasks, using either C++ or Python.

Highlights

• Multi-device support: A single pipeline can manage multiple devices, with automatic host relays for inter-device communication.
• Cross-language support: Written in C++ with official Python bindings.
• Flexible execution: Supports running nodes on the device or the host, with a dedicated beta namespace for experimental features.
• Robust connectivity: Includes built-in reconnection logic and configurable timeouts for device discovery and connection.