The First Intelligent Vision Sensor
Sony's IMX500 is a groundbreaking CMOS sensor that combines image processing and AI on a single chip to reduce infrastructure costs, enhance efficiency, and accelerate edge AI deployment.
The edge of vision AI
Sony’s pioneering innovation, the IMX500, is a one-of-a-kind sensor that combines an image processor and AI processing on the same chip - a pinnacle of advanced engineering at the forefront of vision AI deployment.
IMX500 excels in performance and efficiency, reducing infrastructure and accelerating customer deployment. Its seamless integration with AI tools and services empowers OEM partners to deliver cutting-edge vision AI solutions.
Unlike conventional vision AI approaches, IMX500 does image and AI processing on the chip, eliminating the need for server-side processing. This results in lower latency, reduced bandwidth requirements, and improved power efficiency for vision AI solutions.
One-chip solution
The IMX500 is more than an image sensor. With a powerful DSP and dedicated on-chip SRAM, it enables high-speed edge AI processing without needing the round trip to a server.
Optimized processing & low power
With IMX500, everything is on the chip. Deploy a new AI model and reuse the same powerful hardware for different tasks, so your vision AI system will need less bandwidth and less power.
Cost-effective
IMX500 eliminates the need for high-performance processors or external memory. It’s optimized design reduces bandwidth and power needs, making it cost-effective.
Address privacy concerns
Safeguard privacy in vision applications with IMX500 which only outputs the metadata of processed images – the image data never needs to leave the chip.
IMX500 key specifications
Key specification | ||
|---|---|---|
Key specification Number of effective pixels | 4056(H)×3040(V), approx. 12.3 megapixels | |
Key specification Image size | Diagonal 7.857 mm (1/2.3 type) | |
Key specification Frame rate | Full pixel | 60fps |
Key specification | 4K video | (4056×2288）60fps |
Key specification | 1080p video | 240fps |
Key specification Frame rate | Full/video + AI processing | 30fps |
Key specification | Metadata output | 30fps |
Key specification Main functions | AI processing function, ISP, HDR shooting | |
Key specification Input tensor size | Min | 64(H)×48(V) |
Key specification | Max | 640(H)×480(V) |
Key specification Input data type | int8 or uint8 | |
Key specification Supported machine learning framework | TensorFlow Lite | |
Key specification Memory size | 8388480 bytes for firmware, network weight file, and working memory | |
Key specification Output | MIPI D-PHY 1.2 (4lane)/SPI | |
Key specification Unit cell size | 1.55μm(H)×1.55μm(V) | |
Key specification Sensitivity (F5.6 standard value) | Approx. 250LSB | |
Key specification Sensor saturation signal level (minimum value) | Approx. 9610e- | |
Key specification Power supply | Analog | 2.7V |
Key specification | Digital | 0.84V |
Key specification | Interface | 1.8V |
Key specification Color filter array | Bayer array | |
Key specification Output formats | Image (Bayer RAW), ISP output (YUV/RGB), ROI, metadata |
“Our IMX500 intelligent vision sensor represents a breakthrough in AI processing. It combines the capabilities of a CMOS image sensor with the ability to process AI without any additional onboard compute. As a result, camera manufacturers can deliver smart cameras with a lower bill-of-materials, reduced power consumption and yet run AI models at high frame rates”
- Eita Yanagisawa,
Senior General Manager of System Solution Business Division,
Sony Semiconductor Solutions Corporation
Simplify vision AI with edge devices powered by IMX500
The IMX500's compact design simplifies and streamlines every aspect of edge AI, from development to deployment. On AITRIOS, edge AI devices equipped with the IMX500 intelligent vision sensor redefine the possibilities of vision AI.
