TLDR
- Raspberry Pi has partnered with Hailo to create a $70 AI Kit add-on for the Raspberry Pi 5 microcomputer.
- The AI Kit includes a Hailo-8L AI accelerator module capable of 13 trillion operations per second (TOPS), outperforming some laptop AI chips.
- The AI Kit integrates with the Raspberry Pi’s camera software stack and can run AI tasks such as object detection, pose estimation, and facial recognition.
- The kit aims to make AI more accessible and power-efficient for professionals and enthusiasts alike.
- The software installation process is simple, and users can run AI demos within minutes of setting up the kit.
Raspberry Pi, the beloved single-board computer maker, has recently unveiled a new addition to its lineup: the Raspberry Pi AI Kit. Developed in collaboration with AI chipmaker Hailo, this $70 add-on brings the power of artificial intelligence to the Raspberry Pi 5 microcomputer.
The kit combines the Raspberry Pi M.2 HAT+ with a Hailo-8L AI accelerator module, offering users a cost-effective and energy-efficient solution for integrating high-performance AI functionality into their projects.
At the heart of the AI Kit lies the Hailo-8L module, a compact yet powerful AI accelerator capable of delivering an impressive 13 trillion operations per second (TOPS).
This performance surpasses that of some laptop AI chips, such as AMD’s first-generation XDNA Ryzen 7040-series and Intel’s Meteor Lake processors.
The Hailo-8L connects to the Raspberry Pi 5 via a single-lane PCIe 3.0 connection, ensuring fast and efficient data transfer.
One of the key features of the Raspberry Pi AI Kit is its seamless integration with the Raspberry Pi’s camera software stack.
This enables users to rapidly develop sophisticated AI vision applications that run in real-time with low latency and minimal power consumption.
The kit is compatible with both first-party and third-party cameras, providing flexibility for various use cases.
The AI Kit empowers users to tackle a wide range of AI tasks, including object detection, semantic and instance segmentation, pose estimation, and facial landmarking.
These computationally intensive tasks are handled entirely by the Hailo-8L co-processor, freeing up the Raspberry Pi 5’s CPU to focus on other critical functions.
The kit’s efficient hardware scheduling allows users to run multiple neural networks on a single camera or process data from two cameras concurrently.
To simplify the software aspect of AI development, Raspberry Pi has worked diligently to integrate the camera subsystem with the AI framework.
The rpicam-apps suite of camera applications now includes a post-processing template that enables real-time neural network inferencing within the camera pipeline.
By leveraging the pre-installed Hailo Tappas post-processing libraries, users can create advanced AI-based applications with just a few hundred lines of C++ code.
Getting started with the Raspberry Pi AI Kit is a easy – users need only install a few packages through apt, reboot the system, and they can begin experimenting with the provided AI demos within minutes.