Raspberry Pi, one of the organisations behind some of the cheapest, smallest single-board computers in the business, is ready to show an add-on that would unlock plenty of applications—and yes, since it’s 2024, there is an AI component here. The Raspberry Pi AI Camera features an on-board image sensor with onboard AI processing for $70.
The Sony image sensor—called the IMX500—is used on the AI Camera in conjunction with the RP2040 that features the proprietary on-chip SRAM microcontroller from Raspberry Pi. As with other offerings in the line-up, the RP2040 speaks to the central philosophy of Raspberry Pi: affordability with efficiency.
In other words, AI startups are not about to swap out their Nvidia GPUs with RP2040s for their inference jobs. Hook it up with an image sensor, however, and it becomes an extension module capable of capturing an image and processing it using a variety of widely supported neural network models. 
Onboard processing in the module keeps the computational burden of visual data away from involvement with the host Raspberry Pi. The Raspberry Pi remains free for general use, and there is no need for a companion accelerator. The new module is compatible with all Raspberry Pi computers.
This is not the first camera module that Raspberry Pi has created. It also still sells the Raspberry Pi Camera Module 3, a basic 12-megapixel image sensor from Sony (IMX708) mounted on a small add-on board that can be attached to a Raspberry Pi with a ribbon cable. The policy of the Raspberry Pi Company to keep their offerings in production for a long time guarantees that the Camera Module 3 will remain available for some time at an approximate price of $25.
The AI Camera is exactly 25mm x 24mm, as is the Camera Module 3, though a little thicker owing to the design of the optical sensor. It runs the MobileNet-SSD model—a real-time object detection framework.
By now, you may wonder who would use the Raspberry Pi AI camera.
While it was designed for the tech-savvy and homelab projects, Raspberry Pi has repositioned itself henceforth, targeting companies that do value addition with these gadgets onto their products or use them within their production lines towards some industrial purpose.
When Raspberry Pi went public, it announced that the industrial and embedded sectors accounted for 72% of its sales. That ratio is likely to be even higher for the AI camera.
This can range from smart city environments, where the AI Camera module 3 can be used for monitoring empty parking spots or for the flow of traffic. In industry, the hardware provides basic automated quality assurance by means of objects passing under the camera module.
Companies like the fact that they can manufacture computers and modules in volume with the Raspberry Pi. They certainly had some supply issues post-COVID, but those appear to be behind them now. Companies know they can dependably source the products from Raspberry Pi without delays in their pipeline. One of the reasons Raspberry Pi is comfortable in saying the AI Camera will stay in production until at least January 2028.
