Robust pedestrian detection under deformation using simple boosted features
作者：
Highlights：
• We propose deformation-robust pedestrian detection method using BMF and BDF.
• We incorporate a spatial pyramid pool method that uses multiple sized blocks.
• We use a RealBoost method to train a tree-structured classifier.
• We also apply a region-of-interest method to remove false positives effectively.
• Our detector achieved the smallest log-average miss rates among all state-of-the-art pedestrian methods.
摘要
•We propose deformation-robust pedestrian detection method using BMF and BDF.•We incorporate a spatial pyramid pool method that uses multiple sized blocks.•We use a RealBoost method to train a tree-structured classifier.•We also apply a region-of-interest method to remove false positives effectively.•Our detector achieved the smallest log-average miss rates among all state-of-the-art pedestrian methods.
论文关键词：Regionlet,Pedestrian detection,Selective max pooling,Selective difference pooling,Boosted tree-structured classifier
论文评审过程：Received 28 March 2016, Revised 17 December 2016, Accepted 8 February 2017, Available online 27 February 2017, Version of Record 9 March 2017.
论文官网地址：https://doi.org/10.1016/j.imavis.2017.02.007
