Double Mask R-CNN for Pedestrian Detection in a Crowd

Figure 3

Pseudosemantic segmentation annotation. (a) Original image. (b) Our method. (c) Method of [31].