Raspberry Pi and Sony have collaborated to launch a new AI-powered camera module priced at $70, aiming to simplify the development of edge AI solutions that process visual data. The Raspberry Pi AI Camera integrates Sony’s IMX500 image sensor with an onboard AI accelerator, allowing users to run neural network models directly on the device without the need for additional hardware like GPUs or external accelerators.
The AI Camera is built around a 12.3-megapixel Sony IMX500 Intelligent Vision Sensor and features Raspberry Pi’s RP2040 microcontroller for neural network and firmware management. This combination enables real-time image processing and inference, freeing up the Raspberry Pi’s processor for other tasks.
Key specifications of the AI Camera include:
- Resolution: 4056×3040 at 10 frames per second, and 2028×1520 at 30 frames per second
- Pixel Size: 1.55 µm × 1.55 µm
- Field of View: 78 degrees with manually adjustable focus
- Compatibility: Works with all Raspberry Pi models, including the Raspberry Pi Zero
One of the standout features is the camera’s seamless integration with Raspberry Pi’s existing software stack. The device uses the libcamera framework, allowing developers to retrieve synchronized image frames and neural network outputs effortlessly. This integration makes it easier to build applications like object detection and pose estimation without extensive coding. For instance, running common models like MobileNet SSD for object detection or PoseNet for pose estimation can be achieved with minimal lines of code using libraries like Picamera2.
The AI Camera also supports converting existing neural network models from popular frameworks such as TensorFlow or PyTorch to run efficiently on the device. Sony’s suite of AI tools facilitates this conversion, and a model zoo with optimized models is available for users who want to get started quickly.
While the AI Camera offers substantial benefits, Raspberry Pi also provides the AI Kit (which it launched in June), a more powerful alternative capable of performing thirteen trillion operations per second. The AI Kit, however, is only compatible with the Raspberry Pi 5 and requires a separate camera module. Choosing between the AI Camera and the AI Kit depends on the specific needs of the project, such as performance requirements and hardware compatibility.
The introduction of the AI Camera opens up new possibilities for developers interested in edge AI applications. Potential use cases range from smart city sensors monitoring traffic flows to industrial quality assurance systems. The device’s compact size—measuring 25 x 24 x 11.9 mm—and its compatibility with all Raspberry Pi models make it a versatile tool for both hobbyists and professionals.
If you're interested in exploring the capabilities of the AI Camera, Raspberry Pi has released a Getting Started Guide that includes instructions on hardware installation, software setup, and running example models from the model zoo.
The Raspberry Pi AI Camera is now available for purchase, and the company has committed to keeping it in production until at least January 2028.