Raspberry Pi Unveils AI Kit with Hailo Collaboration

Raspberry Pi has launched a new AI Kit developed in partnership with Hailo, designed to make artificial intelligence (AI) more accessible for developers working on low-power edge applications. Priced at $70, the Raspberry Pi AI Kit is engineered to facilitate the experimentation and implementation of neural networks, AI, and machine learning on the Raspberry Pi 5 platform.
The collaboration between Raspberry Pi and Hailo aims to provide a convenient means for integrating efficient, high-performance inferencing locally into a broad array of applications. The kit enables the construction of complex AI vision applications that operate in real-time, featuring low latency and minimal power consumption. Thanks to the built-in Hailo-8L co-processor, the kit boasts the capability to run neural networks for various tasks, including object detection, semantic and instance segmentation, pose estimation, and facial landmarking entirely on-device.
Offering an inferencing performance of 13 TOPS and a single-lane PCIe 3.0 connection operating at 8Gbps, the kit seamlessly integrates with the Raspberry Pi image software subsystem. It supports both first-party and third-party cameras, as per the available documentation. For users not interested in developing their neural networks, Hailo offers an array of pre-trained models ready for immediate deployment.
The software’s primary focus is on enabling AI-vision applications. It achieves compatibility with cameras through the rpicam-apps suite, which includes a post-processing template for embedding neural network inferencing within the camera pipeline in real-time. The inclusion of Hailo Tappas post-processing libraries allows users to develop AI applications with just a few hundred lines of C++ code.
While Raspberry Pi has showcased several application previews, the kit’s versatility extends beyond the rpicam-apps or Picamera2, incorporating an API within the GStreamer framework and facilitating native Python or C/C++ applications for various use cases, including inference on pre-recorded video files.
For detailed information on the AI Kit and the technical specifications of the Hailo-8L AI accelerator module, interested parties are encouraged to visit the official Raspberry Pi and Hailo product web pages.
The article is penned by Ken Briodagh, a seasoned writer and editor passionate about technology, who has a rich background spanning various roles and industries.