GTC-DC 2019: From Theory to Practice: Computer Vision on Edge Devices for Real-Time Traffic Optimization - Overview
Note: Viewing this video may require joining the NVIDIA Developer Program or login in
GTC-DC 2019: From Theory to Practice: Computer Vision on Edge Devices for Real-Time Traffic Optimization
Yoav Valinsky, NoTraffic
We’ll show how NoTraffic autonomously optimizes traffic lights in real time using multiple sensors. Our system maximizes traffic flow, reduces congestion and carbon dioxide emissions, prioritizes different vehicle types, prevents accidents, and enables cities to implement traffic policies. We’ll demonstrate how we use NVIDIA’s hardware and GPU-accelerated frameworks on our edge devices to operate within the constraints of IoT. By fusing multiple deep networks, we can apply existing computer vision concepts to real world scenarios. We’ll discuss the noisy and biased data that’s in every real-world problem, which is rarely addressed in research papers and datasets. By applying active learning, we constantly optimize our training and data collection pipelines for continuous deployment of new models.