$70 AI Camera Brings On-Device ML to Raspberry Pi Ecosystem

Raspberry Pi unveiled a $70 AI Camera module with onboard processing for industrial and embedded applications in smart sensing and automated visual inspection. The camera module features image capture capabilities with onboard AI processing for edge computing and machine vision applications. Use cases could include smart city sensors for parking management and traffic monitoring, and automated quality control in manufacturing settings. The AI Camera offers enhanced capabilities for projects requiring computer vision and machine learning at the edge.

Sony’s IMX500 image sensor is paired with Raspberry Pi’s RP2040 MCU chip so that the module captures 12.3-megapixel images and performs AI inference tasks directly on the device. Pre-loaded with the MobileNet-SSD object detection model, the camera enables real-time object recognition.

The module offloads AI processing from the Raspberry Pi host computer, leaving the main Raspberry Pi free to perform other operations without additional accelerator hardware. The camera is compatible with existing Raspberry Pi computers, connects via a standard ribbon cable, and measures 25mm x 24mm.

The software includes integration with popular machine learning frameworks. Users can convert PyTorch and TensorFlow models for the device, allowing developers to deploy custom AI models tailored to their application needs.

The AI Camera maintains 30 frames per second for tasks like human pose estimation and object detection at 640×480 resolution. This real-time processing capability makes it suitable for responsive, low-latency applications. Raspberry Pi will keep the AI Camera in production until at least January 2028 for companies looking to incorporate the module into their products or industrial processes. This long-term availability is crucial for businesses that require a stable supply chain for their hardware components.