Raspberry Pi introduces $70 AI camera

Built around the advanced Sony IMX500 image sensor, the AI Camera comes equipped with an integrated AI accelerator, designed to run a variety of popular neural network models with low power consumption and minimal latency. This leaves the Raspberry Pi’s processor free for other tasks, enhancing overall performance.

The AI Camera features a 12 MP Sony IMX500 Intelligent Vision Sensor, capable of capturing high-resolution images at 4056×3040 at 10fps or 2028×1520 at 30fps. Its 1.55 µm × 1.55 µm cell size and 78-degree field of view make it suitable for a wide range of applications, with manually adjustable focus providing added flexibility. Additionally, the camera is powered by an integrated RP2040 microcontroller, managing neural network processing and firmware tasks efficiently.

Compatible with all models of Raspberry Pi, including the Raspberry Pi Zero, the AI Camera connects via regular camera ribbon cables, making it a versatile and accessible option for users of different devices. The camera leverages Sony’s AI tools, enabling existing neural network models built on frameworks such as TensorFlow or PyTorch to be easily converted for use on the AI Camera. Alternatively, users can develop custom models to harness the full potential of the AI accelerator.

When considering whether to opt for the Raspberry Pi AI Camera or the more powerful Raspberry Pi AI Kit, users should note that the AI Kit offers higher theoretical performance and supports a wider range of models but is only compatible with the Raspberry Pi 5. In contrast, the AI Camera is more compact, affordable, and compatible with all Raspberry Pi models, making it an attractive choice for those who prioritise versatility and a lower overall cost.

The Raspberry Pi AI Camera offers excellent acceleration for commonly used models and works seamlessly with the existing Raspberry Pi camera software stack. To help users get started, Raspberry Pi has provided a detailed Getting Started Guide, featuring instructions on hardware setup, software environment configuration, and examples from their neural network model zoo.

Images courtesy of the Raspberry Pi Foundation