Raspberry Pi AI Camera – Sony IMX500 12.3MP Intelligent Vision Sensor

The Raspberry Pi AI Camera combines the Sony IMX500 12.3MP Intelligent Vision Sensor with on-sensor AI processing, allowing neural network inference to be performed directly within the camera system.

This integrated AI capability makes it possible to build advanced computer vision applications without requiring a separate external AI accelerator for many common inference workloads.

Designed to integrate closely with the Raspberry Pi camera software ecosystem, the AI Camera supports libcamera, Picamera2, and rpicam-apps, making it suitable for beginners, developers, researchers, robotics projects, machine vision, and professional edge-AI applications.

Key Features

• Sony IMX500 Intelligent Vision Sensor
• 12.3MP resolution
• Maximum resolution of 4056 × 3040 pixels
• Integrated AI inference acceleration
• Run supported neural network models directly through the camera system
• Reduces the need for a separate AI accelerator for supported vision workloads
• Integrated RP2040 for neural network firmware management
• Support for tensor metadata
• Integration with libcamera
• Integration with Picamera2
• Compatible with rpicam-apps
• Manual adjustable focus
• Integrated IR-cut filter
• Supports Raspberry Pi camera ecosystem
• Suitable for machine learning and edge-AI applications

Integrated AI Processing

The Sony IMX500 combines image sensing with AI processing capabilities, enabling supported neural network inference to be performed close to the image sensor.

This architecture is particularly useful for applications such as:

• Object detection
• Object classification
• People detection
• Security monitoring
• Machine vision
• Robotics
• Smart cameras
• Edge AI
• Industrial monitoring

By processing AI workloads locally, the camera can reduce the amount of image data that needs to be processed by the Raspberry Pi CPU.

Raspberry Pi Software Integration

The Raspberry Pi AI Camera is tightly integrated with the Raspberry Pi camera software stack.

It supports tensor metadata through:

• libcamera
• Picamera2
• rpicam-apps

This allows developers to access AI inference results alongside captured image data and build intelligent camera applications using familiar Raspberry Pi development tools.

Technical Specifications

• Image Sensor: Sony IMX500
• Resolution: 12.3MP
• Maximum Resolution: 4056 × 3040 pixels
• Sensor Size: 7.857mm (Type 1/2.3)
• Pixel Size: 1.55 × 1.55µm
• IR-Cut Filter: Integrated
• Infrared Sensitive: No
• Focus System: Manual adjustable focus
• Focus Range: 20cm to infinity
• Focal Length: 4.74mm
• Horizontal Field of View: Approx. 66° ±3°
• Vertical Field of View: Approx. 52.3° ±3°
• Focal Ratio: F1.79
• AI Management Controller: RP2040
• Maximum Input Tensor Size: 640 × 640
• Input Data Type: int8 or uint8
• Memory: 8,388,480 bytes for firmware, network weights, and working memory
• Camera Output: Bayer RAW10, ISP output (YUV/RGB), ROI and metadata
• Operating Temperature: 0°C to +50°C
• Camera Dimensions: 25 × 24 × 11.9mm
• Ribbon Cable Length: 200mm

Frame Rates

The camera supports:

• 2028 × 1520: 10-bit, up to 30fps with 2 × 2 binning
• 4056 × 3040: 10-bit, up to 10fps at full resolution

Camera Connection

The camera supports Raspberry Pi camera connections using compatible FPC ribbon cables:

• 15-pin, 1mm-pitch FPC
• 22-pin, 0.5mm-pitch FPC

The required cable depends on the Raspberry Pi model being used.

Typical Applications

• AI computer vision
• Object detection
• Image classification
• Intruder detection
• Smart security cameras
• Robotics vision
• Machine learning
• Industrial machine vision
• Smart home systems
• IoT applications
• People detection
• Automated inspection
• Edge AI
• Educational AI projects
• Research and development

Long-Term Availability

The Raspberry Pi AI Camera is planned to remain in production until at least January 2028, making it suitable for projects that require longer-term product availability.

Compatibility Note

The Raspberry Pi AI Camera is designed for use with Raspberry Pi computers that support the Raspberry Pi camera interface. The required 15-pin or 22-pin ribbon cable depends on the Raspberry Pi model, so the correct cable should be selected for the target board.