Dynamic Error-bounded Lossy Compression (EBLC) to Reduce the Bandwidth Requirement for Real-time Vision-based Pedestrian Safety Applications

As camera quality improves and their deployment moves to areas with limited\nbandwidth, communication bottlenecks can impair real-time constraints of an ITS\napplication, such as video-based real-time pedestrian detection. Video\ncompression reduces the bandwidth requirement to transmit the video but\ndegrades the video quality. As the quality level of the video decreases, it\nresults in the corresponding decreases in the accuracy of the vision-based\npedestrian detection model. Furthermore, environmental conditions (e.g., rain\nand darkness) alter the compression ratio and can make maintaining a high\npedestrian detection accuracy more difficult. The objective of this study is to\ndevelop a real-time error-bounded lossy compression (EBLC) strategy to\ndynamically change the video compression level depending on different\nenvironmental conditions in order to maintain a high pedestrian detection\naccuracy. We conduct a case study to show the efficacy of our dynamic EBLC\nstrategy for real-time vision-based pedestrian detection under adverse\nenvironmental conditions. Our strategy selects the error tolerances dynamically\nfor lossy compression that can maintain a high detection accuracy across a\nrepresentative set of environmental conditions. Analyses reveal that our\nstrategy increases pedestrian detection accuracy up to 14% and reduces the\ncommunication bandwidth up to 14x for adverse environmental conditions compared\nto the same conditions but without our dynamic EBLC strategy. Our dynamic EBLC\nstrategy is independent of detection models and environmental conditions\nallowing other detection models and environmental conditions to be easily\nincorporated in our strategy.\n