Abstract
This paper presents a novel end-to-end system for pedestrian detection usingDynamic Vision Sensors (DVSs). We target applications where multiple sensorstransmit data to a local processing unit, which executes a detection algorithm.Our system is composed of (i) a near-chip event filter that compresses anddenoises the event stream from the DVS, and (ii) a Binary Neural Network (BNN)detection module that runs on a low-computation edge computing device (in ourcase a STM32F4 microcontroller). We present the system architecture and providean end-to-end implementation for pedestrian detection in an office environment.Our implementation reduces transmission size by up to 99.6% compared totransmitting the raw event stream. The average packet size in our system isonly 1397 bits, while 307.2 kb are required to send an uncompressed DVS timewindow. Our detector is able to perform a detection every 450 ms, with anoverall testing F1 score of 83%. The low bandwidth and energy properties of oursystem make it ideal for IoT applications.