Raspberry Pi AI Camera

Raspberry Pi AI Camera: Smart computer vision, without external accelerators

Turn your Raspberry Pi into a complete computer vision system in just a few minutes, hassle-free, with the Raspberry Pi AI Camera, based on the Sony IMX500 sensor. A compact and powerful module, perfect for computer vision projects without the need for external AI hardware.

The Sony IMX500 sensor combines a 12-megapixel CMOS camera with an integrated AI accelerator. This allows you to perform neural processing directly on the camera, freeing up your Raspberry Pi’s processor for other tasks.

The Raspberry Pi AI Camera is ideal for advanced computer vision projects, such as:

• Visual recognition and object recognition projects
• Pose estimation
• Real-time video analysis
• Robotics projects
• Smart automation and surveillance (Video surveillance with edge processing)
• AI prototypes for industrial or educational applications

The Raspberry Pi AI Camera is plug-and-play with full software support, so it’s easy to use even for beginners, thanks to its compatibility with:
libcamera and Picamera2
rpicam-apps application suite
AI data (tensor metadata) is automatically attached to images and videos, ready for use in your applications.

The AI Camera works with all Raspberry Pi models equipped with a CSI port, including:
Raspberry Pi 5
Raspberry Pi 4, 3, Zero and CM (with CSI)

The board dimensions and mounting holes are identical to those of the Camera Module 3, but with greater depth to accommodate the advanced sensor.

The pack contains:

• 1x Raspberry Pi AI Camera
• 1x 15-pin CSI connection cable (for Pi 2/3/4)
• 1x 22-pin CSI connection cable (for Pi Zero/Pi 5)
• 1x Camera Focus Adjustment Tool

Note: Raspberry Pi not included.

Key features of the Raspberry Pi AI Camera

• Sony IMX500 sensor with integrated AI acceleration
• On-board AI inference (no need for HATs or external modules)
• Native support in official Raspberry Pi libraries and apps
• Tensor metadata embedded in every captured image/video
• Compatible with all Raspberry Pi models equipped with CSI

Technical Specifications of the Raspberry Pi AI Camera

• Sensor: Sony IMX500
• Resolution: 12.3 megapixels
• Sensor size: 7.857 mm (1/2.3-type)
• Pixel size: 1.55 μm × 1.55 μm
• Image resolution (horizontal/vertical): 4056 × 3040 pixels
• IR cut filter: Integrated
• Focusing system: Manual adjustment
• Focusing distance: 20 cm to ∞
• Focal length: 4.74 mm
• Horizontal field of view: 66° ±3
• Vertical field of view: 52.3° ±3
• Aperture (F-stop): F1.79
• Infrared sensitive: No
• Available outputs
  Image (Bayer RAW10)
  ISP output (YUV/RGB)
  ROI (Region of Interest)
  Metadata (Tensor metadata)
• AI and performance
  Maximum input tensor size: 640(H) × 640(V)
• Input data type: int8 or uint8
• Internal memory: 8,388,480 bytes for firmware, network weights and operational memory
• Frame rate
  2×2 binning: 2028 × 1520 at 10 bit, 30 fps
  Full resolution: 4056 × 3040 at 10-bit, 10 fps
• Module dimensions: 25 × 24 × 11.9 mm
• Flat cable length: 200 mm
• Connector type:
  15 × 1 mm FPC
  22 × 0.5 mm FPC
• Operating environment
  Operating temperature: 0°C to +50°C

Useful resources – Raspberry Pi AI Camera:

• Getting Started with Raspberry Pi AI Camera
• Raspberry Pi AI Camera Product Brief
• Model Zoo
• Sony AITRIOS Developer site for AI Camera
• Sony Raspberry Pi AI Camera IMX500 Packager Manual
• Sony Raspberry Pi AI Camera IMX500 Converter Manual
• Raspberry Pi AI Camera Launch Blog Post
• Libcamera
• Picamera2
• Rrpicam-apps