Intelligent Video Ingestion for Real-time Traffic Monitoring

As an indispensable part of modern critical infrastructures, cameras deployed at strategic places and prime junctions in intelligent transportation system can help operators observing traffic flow, identifying any emergency situation, or making decisions regarding road congestion without arriving on the scene. However, these are usually equipped with heterogeneous turbulent networks, real-time smooth playback monitoring videos high quality a grand challenge. In this article, we propose lightweight Deep Reinforcement Learning-based approach, namely, sRC-C (smart bitRate Control Continuous action space) , to enhance by adjusting video bitrate adaptively. Distinguished from existing approaches, overcome bias incurred deterministic discretization candidate bitrates more fine-grained control continuous space, thus significantly improving Quality-of-Service (QoS). With carefully designed state space neural network model, be implemented scarce resources support live streaming low inference time. Extensive experiments show that reduce frame loss counts hold time 24% 15.5%, respectively, even comparable bandwidth utilization. Meanwhile, compared the-state-of-art improve QoS 30.4%.