OSS-OCL: Occlusion Scenario Simulation and Occluded-edge Concentrated Learning for pedestrian detection

Keqi Lu, Chao Zhu, Mengyin Liu, Xu-Cheng Yin

Abstract: Highlights•Occlusion Scenario Simulation is proposed to simulate the most common occlusion scenario.•Occluded-edge Concentrated Learning is proposed to guide the detector to focus on the occluded-edge.•Our proposed method outperforms some SOTA methods on various popular pedestrian detection datasets.