The Raspberry Pi AI Camera, priced at $70, is equipped with the Sony IMX500 Intelligent Vision Sensor, featuring a 12.3-megapixel image sensor and an AI-inferencing accelerator.

At its core, the camera utilizes an RP2040 processor, allowing it to handle AI tasks independently without overloading the Raspberry Pi's CPU.

This device can recognize objects, colors, shapes, and orientation, making it suitable for various applications.

The AI Camera comes preloaded with the MobileNetSSD model for real-time object detection and will be supported by Raspberry Pi until at least January 2028.

Raspberry Pi has designed the camera to enhance usability for beginners while providing advanced functionalities for experienced users, including support for tensor metadata.

Users have reported an easy setup process and low CPU usage during AI processing, although more complex tasks may increase CPU load.

The camera supports neural network frame firmware and can augment images with tensor metadata for further processing using tools like rpicam-apps or Picamera2.

It features built-in AI processing, enabling users to create advanced AI solutions for visual data processing, including generating new images from captured data.

The camera connects to the Raspberry Pi via a flat ribbon cable to the CSI port, making it compatible with all Raspberry Pi models.

This product marks the first collaborative technology release since Sony's investment in Raspberry Pi last year, highlighting a significant partnership.

Onboard processing allows the Raspberry Pi to remain free for other operations, eliminating the need for a separate accelerator.

The project is open source, with detailed information available on GitHub, catering to DIY enthusiasts and developers.