DESCRIPTION
The Raspberry Pi AI Camera brings on-sensor artificial intelligence to your Raspberry Pi, built around the 12.3MP Sony IMX500 intelligent vision sensor with its integrated RP2040 co-processor. Unlike a standard camera module, it runs neural network inference directly on the sensor itself, making it ideal for makers, students, and Industry 4.0 prototypers who want real-time object detection, pose estimation, or classification without loading down the Pi's own CPU.
Because the neural network runs on-chip, results stream out as metadata alongside (or instead of) full video frames, giving you fast, low-power AI vision that leaves the Raspberry Pi free to handle the rest of your application.
FEATURES
- 12.3MP Sony IMX500 intelligent vision sensor with integrated RP2040 for on-sensor neural network firmware management
- On-sensor AI inference frees up the Raspberry Pi's CPU and GPU for other tasks
- F1.79 lens with manual/mechanical adjustable focus and a wide 78.3° (±3°) field of view
- Captures up to 4056x3040 at 10fps (full resolution) or 2028x1520 at 30fps (2x2 binned), plus 1080p30 video
- Standard MIPI CSI-2 camera connector — works with the same camera port as other Raspberry Pi camera modules
- Compatible with Raspberry Pi 5, Raspberry Pi 4 Model B, Raspberry Pi 3 Model B+, and Raspberry Pi Zero 2 W
- Ships with pre-loaded neural network models — including MobileNet SSD object detection and PoseNet pose estimation — ready to run out of the box
- Supports custom-trained neural networks via the imx500-tools model packaging toolkit
- Compact 25 x 24 x 11.9mm module with a standard 4-hole mounting pattern
APPLICATIONS
- Real-time object detection and tracking for robotics projects
- Smart home and security cameras with on-device person or motion detection
- People counting and pose estimation for STEM and research projects
- Low-latency machine vision for Industry 4.0 and automation prototypes
- Wildlife or maker-community camera traps that classify subjects on-device
- Educational introduction to edge AI and computer vision on Raspberry Pi
TECHNICAL DETAILS
- Sensor: Sony IMX500 intelligent vision sensor (12.3 MP) with integrated RP2040 co-processor for on-sensor neural network firmware management
- Full Resolution: 4056x3040, 10-bit, 10fps
- 2x2 Binned Resolution: 2028x1520, 10-bit, 30fps
- Video Mode: 1080p30
- Lens: F1.79 focal ratio, manual/mechanical adjustable focus
- Field of View: 78.3° (±3°) diagonal
- Sensor Size: 7.857mm; Pixel Size: 1.55µm x 1.55µm
- Module Dimensions: 25 x 24 x 11.9mm
- Interface: MIPI CSI-2 (standard Raspberry Pi camera connector)
- Compatibility: Raspberry Pi 5, Raspberry Pi 4 Model B, Raspberry Pi 3 Model B+, Raspberry Pi Zero 2 W (works with any Raspberry Pi camera connector using the appropriate cable)
- Software: Requires the latest Raspberry Pi OS software stack; works with rpicam-apps, libcamera, and Picamera2 via the imx500-all firmware and model package
- Pre-loaded Neural Network Models: MobileNet SSD (object detection), PoseNet (pose estimation), with YOLOv8 supported via Picamera2
SHIPPING LIST
- 1x Raspberry Pi AI Camera (IMX500)
- 1x Raspberry Pi Camera Cable (Standard – Mini, 200mm)
