Fast Head-Shoulder Proposal for Scare-Aware Pedestrian Detection
Tianrui Liu, Tania Stathaki · 2017
In this paper, we propose a scare-aware pedestrian detection method assisted by a fast head-shoulder detection process to pre-estimate the candidate regions that may contain target pedestrians. Based on the observation that human head-shoulder regions share relatively robust features, we propose a head-shoulder detector using six aggregated feature channels. Motivated by fact that visual appearance of the large-size and small-size pedestrians are significantly different, we apply a scale-aware detection mechanism. Two detectors are used, each focuses on the detection of either large-size or small-size pedestrians. The detection results from two detectors are weighted combined so that final detection results can be reinforced by a proper detector with respect to the size of the candidate pedestrian.