Pedestrian Gender Detection Based on Mask R-CNN

Abstract

Based on Mask R-CNN, one of the most significant models for instance segmentation, we present our pedestrian gender detection algorithm in this paper. First of all, we verify the effectiveness of Mask R-CNN in pedestrian detection. On this basis, we design our network by merging Mask R-CNN with a gender recognition branch. And due to the lack of current datasets, we not only present a method to train our model, but also build a shopping mall dataset to test our model. The application of our detection model to our small dataset is of great significance to the business planning and the security maintenance of the entertainment center.