Evaluating State-Of-The-Art Object Detector on Challenging Traffic Light Data

Morten B. Jensen, Kamal Nasrollahi, Thomas B. Moeslund; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2017, pp. 9-15

Abstract


Traffic light detection (TLD) is a vital part of both intelligent vehicles and driving assistance systems (DAS). hard to determine the exact performance of a given method. In this paper we apply the state-of-the-art, real-time object detection system You Only Look Once, (YOLO) on the public LISA Traffic Light dataset available through the VIVA-challenge, which contain a high number of annotated traffic lights, captured in varying light and weather conditions. The YOLO object detector achieves an AUC of impressively 90.49 % for daysequence1, which is an improvement of 50.32 % compared to the latest ACF entry in the VIVA-challenge. Using the exact same training configuration as the ACF detector, the YOLO detector reaches an AUC of 58.3 %, which is in an increase of 18.13 %.

Related Material


[pdf]
[bibtex]
@InProceedings{Jensen_2017_CVPR_Workshops,
author = {Jensen, Morten B. and Nasrollahi, Kamal and Moeslund, Thomas B.},
title = {Evaluating State-Of-The-Art Object Detector on Challenging Traffic Light Data},
booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {July},
year = {2017}
}