Raspberry Pi, which sells small, inexpensive single-board computers, has released add-ons that open up several use cases. And, sure, it's 2024, so there's an AI perspective as well. The image sensor, called the Raspberry Pi AI Camera, has onboard AI processing and will cost $70.
More technically speaking, the AI camera is based on a Sony image sensor (IMX500) combined with Raspberry Pi's own microcontroller chip, the RP2040, with on-chip SRAM. Like the rest of the lineup, the RP2040 follows the overall Raspberry Pi philosophy of being cheap, yet efficient.
In other words, AI startups aren't looking to replace Nvidia GPUs with RP2040 chips for inference. But when combined with an image sensor, you get an enhanced module that can capture images and process those images through common neural network models.
As an added benefit, the camera module's onboard processing makes the host Raspberry Pi independent of visual data processing. The Raspberry Pi is free to perform other operations, so there is no need to add another accelerator. The new module is compatible with all Raspberry Pi computers.
This is not the first camera module for Raspberry Pi. The company still sells the Raspberry Pi Camera Module 3. It's a simple 12-megapixel image sensor (IMX708) from Sony attached to a small add-on board that pairs with a Raspberry Pi via a ribbon cable. Since Raspberry Pi has promised to remain in production for many years, the Camera Module 3 will continue to be available for around $25.
The AI camera has the same size as Camera Module 3 (25mm x 24mm) but is slightly thicker due to the construction of the optical sensor. It comes preloaded with the MobileNet-SSD model, an object detection model that can run in real time.
At this point, you may be wondering who would use the Raspberry Pi AI camera. Although this small computer was originally designed for tech enthusiasts and home lab projects, Raspberry Pi now offers most of its devices to companies that use Raspberry Pi devices in their own products, or on assembly lines for internal industrial use cases. We sell it to companies who use it as part of their products.
When Raspberry Pi became a public company, it reported that the industrial and embedded segment accounted for 72% of its sales. With AI cameras, that percentage is likely to be even higher.
You can imagine companies using AI camera modules for smart city sensors that detect empty parking lots or track traffic flow, for example. In an industrial environment, this hardware can be used for basic automatic quality assurance of objects passing under the camera module.
Companies like using Raspberry Pi because it allows them to produce computers and modules at scale. We were facing post-Covid supply constraints, but those seem to have been resolved. Companies know they can reliably source Raspberry Pi products without suffering production pipeline delays. This is also one of the reasons why Raspberry Pi promises that its AI camera will be in production until at least January 2028.
Image credit: Romain Dilet / TechCrunch
Source link