Robust Real-Time Pedestrian Detection on Embedded Devices

TL;DR

This paper presents a robust pedestrian detection framework optimized for real-time performance on embedded devices, combining multi-region detection and temporal-spatial analysis, achieving high accuracy and winning a challenge.

Contribution

It introduces a novel framework that enhances pedestrian detection accuracy and speed on embedded systems using Yolo-v3 and multi-region analysis, suitable for various hardware.

Findings

Achieved second place in CVPR 2019 ERTI Challenge

Demonstrated high accuracy on established datasets

Operates in real-time on Nvidia Jetson TX2

Abstract

Detection of pedestrians on embedded devices, such as those on-board of robots and drones, has many applications including road intersection monitoring, security, crowd monitoring and surveillance, to name a few. However, the problem can be challenging due to continuously-changing camera viewpoint and varying object appearances as well as the need for lightweight algorithms suitable for embedded systems. This paper proposes a robust framework for pedestrian detection in many footages. The framework performs fine and coarse detections on different image regions and exploits temporal and spatial characteristics to attain enhanced accuracy and real time performance on embedded boards. The framework uses the Yolo-v3 object detection [1] as its backbone detector and runs on the Nvidia Jetson TX2 embedded board, however other detectors and/or boards can be used as well. The performance of the framework is demonstrated on two established datasets and its achievement of the second place in CVPR 2019 Embedded Real-Time Inference (ERTI) Challenge.