depthai - Flexible AI Network Loading and Video Recording

DepthAI Project Overview

DepthAI is an innovative project designed to provide flexible and comprehensive solutions for a variety of artificial intelligence and depth-sensing applications. This project is open-source and equipped with accessible tools for developers to harness the power of machine learning and computer vision. It enables application of various AI models, facilitates easy integration of complex functionality, and is aimed at rapid prototyping and experimentation.

Key Features

Modular Architecture: The project offers the ability to load different neural networks and create custom pipelines. This flexibility allows developers to tailor DepthAI for specific use cases.
Video Recording: Capture and record video streams with ease using built-in functionalities, providing ways to store and process video data efficiently.
Comprehensive Documentation: Guided tutorials and documentation available at Luxonis' official site for getting started and understanding detailed workings of DepthAI.

Installation

Before using DepthAI, certain setup steps ensure all necessary dependencies are installed on your system:

Clone the Repository: Clone the DepthAI repository from GitHub to use its features and demo programs.
```
git clone --recursive https://github.com/luxonis/depthai.git
```
Install Dependencies:
- Run a one-time installation script to install required system packages.
```
$ sudo curl -fL https://docs.luxonis.com/install_dependencies.sh | bash
```
- Install Python dependencies to ensure compatibility with the demo updates.
```
$ python3 install_requirements.py
```

Using Docker

DepthAI applications can also be executed within Docker containers, providing virtualization benefits:

Running DepthAI Programs: Utilize Docker to run DepthAI demos, allowing easy network and device access from the container.

docker run --privileged -v /dev/bus/usb:/dev/bus/usb --device-cgroup-rule='c 189:* rmw' -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --network host --rm -i -t luxonis/depthai:latest python3 /depthai/depthai_demo.py

Interactive and Command Line Usage

DepthAI offers two main interfaces for user interaction:

QT GUI: A graphical interface for exploring various features interactively.
Command Line: Execute specific models and functionalities directly from the command line with options to tweak performance using parameters.

Applications

DepthAI supports a variety of real-world applications through its apps:

UVC App: Turns OAK cameras into webcams, useful for standard webcam applications.
Record App: Allows synchronized recording of video streams, offering options for encoding formats for depth perception tasks.

Supported AI Models

A plethora of AI models are supported natively, facilitating quick deployment and testing. Some key models include:

Face Detection (ADAS, Retail)
Human Pose Estimation
Various Object Detection Models (e.g., MobileNet-SSD, YOLO)

For users wanting to run their own custom models, comprehensive guidance is available in the documentation.

Usage Statistics and Feedback

In effort to improve DepthAI, anonymous usage statistics are collected, which can be disabled if desired. User feedback is highly valued, and channels like Discord, forums, and direct support email are available for reporting issues and suggestions.

DepthAI is continually evolving, driven by community input and the pursuit of advancing capabilities in AI and edge computing technologies.