Pedestrian detection in underground mines via parallel feature transfer network
作者：
Highlights：
• The paper presents a parallel feature transfer network (PftNet) based detector to tackle the problem of pedestrian detection in underground mines.
• The paper carries out several experiments on UMP2018 which is composed of 18,450 pedestrian pictures in underground mines and annotated by ourselves, as well as INRIA and ETH dataset.
• The experimental results demonstrate that PftNet achieves the state-of-the-art detection efficiency with high accuracy, which is significant to realizing unmanned driving systems in mines.
摘要
•The paper presents a parallel feature transfer network (PftNet) based detector to tackle the problem of pedestrian detection in underground mines.•The paper carries out several experiments on UMP2018 which is composed of 18,450 pedestrian pictures in underground mines and annotated by ourselves, as well as INRIA and ETH dataset.•The experimental results demonstrate that PftNet achieves the state-of-the-art detection efficiency with high accuracy, which is significant to realizing unmanned driving systems in mines.
论文关键词：Pedestrian detection,Underground mine,Deep learning network,Parallel feature transfer,Gated unit,Unmanned driving
论文评审过程：Received 8 May 2019, Revised 22 September 2019, Accepted 1 January 2020, Available online 12 February 2020, Version of Record 18 February 2020.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107195
