Raspberry Pi takes a shot at AI with a camera and on-device processing

October 1

Raspberry Pi has joined the world of AI with a new camera that processes visual data with AI models on the device. The $70 Raspberry Pi AI Camera was built in collaboration with Sony Semiconductor Solutions Corporation (SSS).

The camera embeds Sony’s IMX500 intelligent vision sensor to process images with AI on the chip. That means there is no need for high-end GPUs or accelerators. It’s compatible with Raspberry Pi’s single-board computers, making it relatively cheap and easy to make AI tools that rely on images.

The AI Camera has around 12.3 megapixels and can film at ten frames per second in 4056 x 3040 or 40 fps at 2028 x 1520. The neural network firmware onboard lets the AI image processing run on the camera. This lets the Raspberry handle other elements of an integrated device, opening up options for Raspberry Pi users who don’t want to use the Raspberry Pi AI Kit alone.

Raspberry Pi AI Camera – Object detection demo – YouTube

Watch On

Raspberry AI

The on-chip AI processing reduces the burden on the host device and enables real-time data processing without requiring cloud-based computing power. Processing data locally makes for faster, more efficient AI tools. Because it’s compatible with much of the Raspberry Pi ecosystem, users can still use hardware and software libraries they are familiar with and enjoy using. For instance, the AI Camera is compatible with popular libraries such as lib-camera and Picamera2.

“AI-based image processing is becoming an attractive tool for developers around the world,” said Raspberry Pi Ltd. CEO Eben Upton. “Together with our longstanding image sensor partner Sony Semiconductor Solutions, we have developed the Raspberry Pi AI Camera, incorporating Sony’s image sensor expertise. We look forward to seeing what our community members are able to achieve using the power of the Raspberry Pi AI Camera.”