Edge computing is gaining ground in the vision industry. However, it is struggling with bandwidth and latency limitations when it comes to high-resolution, high-speeda and/or complex multi-camera systems. This presentation explores the bandwidth bottlenecks and technical challenges and investigates technical solutions to implement high-bandwidth, low-latency imaging with AI processing on the edge.

Advanced vision systems rely on one or multiple high-resolution, high-speed cameras. These capture and deliver multiple gigapixels per second that need to be processed and/or compressed in real-time. While camera sensor technology and supporting interfaces are quicky evolving, traditional PCIe/CPU/GPU-based PC-architectures are lacking the required performance to capture, process, and store images at such data […]