Robust Real-Time Pedestrian Detection on Embedded Devices
This work provides a practical solution for real-time pedestrian detection on resource-constrained embedded systems, benefiting applications in robotics, surveillance, and crowd monitoring.
This paper addresses the challenge of real-time pedestrian detection on embedded devices, which is crucial for applications like robot and drone navigation. The proposed framework, built on Yolo-v3, achieves enhanced accuracy and real-time performance by employing fine and coarse detections on different image regions and leveraging temporal and spatial characteristics. It secured second place in the CVPR 2019 Embedded Real-Time Inference (ERTI) Challenge.
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.