Accurate traffic light detection using deep neural network with focal regression loss
作者：
Highlights：
• We propose a method using a DNN to detect small TLs in images captured by cameras.
• Our TL detector has a DNN architecture of encoder-decoder with focal regression loss.
• The focal regression loss reduces loss of well-regressed easy examples.
• We propose freestyle anchor boxes placed at arbitrary locations in a grid cell.
• Our TL detector outperforms the existing other detectors with higher mAP and AUC.
摘要
•We propose a method using a DNN to detect small TLs in images captured by cameras.•Our TL detector has a DNN architecture of encoder-decoder with focal regression loss.•The focal regression loss reduces loss of well-regressed easy examples.•We propose freestyle anchor boxes placed at arbitrary locations in a grid cell.•Our TL detector outperforms the existing other detectors with higher mAP and AUC.
论文关键词：Advanced driving assistance system,Traffic light detection,Small object detection,Deep neural network,Focal regression loss,Freestyle anchor box
论文评审过程：Received 5 April 2019, Accepted 16 April 2019, Available online 1 May 2019, Version of Record 19 May 2019.
论文官网地址：https://doi.org/10.1016/j.imavis.2019.04.003
