Raspberry Pi is entering the artificial intelligence (AI) arena, first with a Hailo 8L AI board coupled to a M.2 HAT+, and now with a brand new camera designed for AI applications and neural network models.
See also: How to add AI superpowers to your Raspberry Pi
The camera can be used for a variety of AI workloads, from image classifications and object detection to more complicated AI assignments such as human pose detection and semantic image segmentation.
Raspberry Pi AI camera tech specs
- Sensor: 12.3 MP Sony IMX500 Intelligent Vision Sensor complete with built-in neural network accelerator
- Framerates:
2×2 binned: 2028×1520 10-bit 30fps
Full resolution: 4056×3040 10-bit 10fps - Sensor size: 7.857 mm
- Pixel size: 1.55 × 1.55 μm
- Field of view: 78.3 (±3) degrees with manual/mechanical adjustable focus
- Aperture: F1.79
- Size: 25 × 24 × 11.9 mm module dimensions
- On-board processor: Integrated RP2040 for neural network firmware management
- Compatibility: Works with all Raspberry Pi models, using our standard camera connector cable
While I’ve used the standard Raspberry Pi camera for AI workloads, getting a camera that’s been designed from the ground up for the task legitimizes the Raspberry Pi as a true AI workhorse. The camera is built around Sony’s IMX500 vision sensor, the world’s first Intelligent Vision Sensor to feature edge processing. The IMX500’s stacked sensor structure combines an image sensor with digital signal processing along with a dedicated on-chip SRAM that allows for on-board high-speed edge AI processing.
Also: Meet the most affordable Raspberry Pi 5 ever
Not only is the chip a low-power, low-latency solution, but it also helps to improve privacy because the image data doesn’t have to leave the chip, only the metadata.
The camera fits any Raspberry Pi, making use of the camera connector. There’s a bit of an installation process, but it’s not too painful by Raspberry Pi standards.
Also: Don’t buy a Raspberry Pi 5 without also buying this amazing accessory
The AI camera is priced at $70, and right now, as is usual with newly-released Raspberry Pi hardware, availability is patchy. If you want one, my advice is to keep watching the usual suspects — Sparkfun, Vilros, PiShop.us, and CanaKit.
