Edge AI Camera with MitySOM-AM62A
Deploy Artificial Intelligence at the Edge with the MitySOM-AM62A embedded vision camera

Overview

Critical Link’s Edge AI Camera leverages the MitySOM-AM62A production-ready system on module. The camera is a flexible, scalable solution for deploying vision-based AI applications at the edge. It utilizes a Texas Instruments Sitara AM62Ax processor, featuring a powerful combination of AI-target resources: quad core ARM Cortex-A53s, a Cortex-R5F real time co-processor, a deep learning accelerator with floating point DSP up to 40 GFLOPS, and a Matrix Multiple Accelerator that supports up to 2 TOPS. Developers gain access to readily available AI models, accelerating hardware and application development.

The MitySOM-AM62A Edge AI Camera gives engineers a powerful, flexible solution for deploying vision-based AI applications at the edge, that is also cost-effective and suitable for long term production. Engineers don’t have to choose between off-the-shelf imagers that will have to be redesigned every few years, or high performance compute solutions that break the budget. Customers can feel secure designing the MitySOM-AM62A Edge AI Camera into their next product, knowing they can achieve first-to-market advantage and have confidence that it will last as long as their design.

Demo units are currently available for this platform. To arrange a demo, contact Critical Link at info@criticallink.com, or contact your local manufacturers representative.

Specifications

The Edge AI Camera utilizes Critical Link’s MitySOM-AM62A system on module with an application-ready carrier board that features:

• Dual Power Options
  • Power Over Ethernet (device)
  • USB-C power
  • Unit will default to POE power if both are provided.
• Video Streaming
  • Micro-HDMI connector to 1080P HD
  • Active H.264 encode and RSTP / network streaming via Gigabit Ethernet
• Camera Interface
  • Up to x4 lanes of MIPI CSI
  • Raspberry Pi Zero 22 pin camera interface
    • IMX219 Based Raspberry Pi II camera
• Image Signal Processing
  • Leverages Full ISP / tuning stack for TI AM62Ax processor
    • Active Debayering with edge enhancement
    • HDR color mapping / scaling
    • warp transforms (e.g., rotation, Lens distortion correction) and resizing
    • White Balance
• AM62A7 C7x256v DSP w/ MMA accelerated Deep Learning and Inferencing engine
  • Supports object detection, facial recognition, segmentation

Click on the images to view the block diagram and layout of the Edge AI Camera:

MitySOM-AM62A Edge AI Camera Demos

Our popular defect detection demonstration uses artificial intelligence at the edge with our Texas Instruments AM62Ax-based SOM. The demo utilizes the yolox-nano-lite-defect-detection model running on the AM62A C7x-MMA accelerator, delivering 65x faster performance vs running on the ARM A53 quad cores. A live video is displayed on the screen with visual indicators and statistics showing total parts, defect rate, and other analytics. Utilizing Edge AI Studio simplifies model training and deployment for new types of parts and materials.

Featured Hardware: MitySOM-AM62A System on Module powers Critical Link’s Edge AI Camera

Example Applications: Factory Automation, Production Yield/Quality Assurance

Contact info@criticallink.com for more information

This demo utilizes the YOLOX-S-Lite machine learning model for real-time vision-based people tracking with statistical insights such as total visitors, current occupancy, and visit duration. It also offers heatmap: highlighting frequently visited areas.

Featured Hardware: MitySOM-AM62A System on Module powering Critical Link’s Edge AI Camera

Example Applications: Retail, Building Automation, Safety/Security

Contact info@criticallink.com for more information

Related Products

MitySOM-AM62A System on Module

MitySOM-AM62 / AM62A / AM62P Development Kit