Video Analytics: Computer Vision

VIDEO ANALYSIS IS COMMONLY USED IN RETAIL, INDUSTRIAL AUTOMATION, SECURITY, HEALTH CARE, AND AUTOMOTIVE

Video analytics with deep learning and pattern recognition automate application tasks that previously needed human involvement.  Video applications that are implementing deep learning include loss prevention, customer traffic and inventory management in stores, defect detection in factories, automatic license detection,  smart parking, and traffic congestion to name a few.  

Deep-AI’s unique, integrated ML training and inference solutions enables deep learning and processing to be done at the edge locations, where frequent updating and retraining of models in real-time needs to be done. Customers save significant CapEx and OpEx as they do not need to use expensive GPU-based training servers in the Cloud or the Datacenter.

Deep-AI has partnered with Xilinx to offer integrated ML training and inference as part of Xilinx’s Smart World AI Video Analytics Solutions.  

Deep-AI’s solution seamlessly integrates in Xilinx’s Video Machine-learning Streaming Server (VMSS) framework.  VMSS targets AI applications such as facial recognition, virtual fencing, crowd statistics, traffic monitoring, and object detection.

Deep-AI provides integrated ML training and inference on the VMSS server, which processes video streams from many cameras.

 

  • Same System can be used for:
  • Fraud Detection/Loss Prevention
  • Inventory Management
  • Supported Neural Networks:
  • CNNs for imaging applications including all popular neural layers, convolutions, max/average pooling, residual shortcut, batch norm
  • Resnet and Mobilenet for Classification
  • More to come