Abstract:It is a challenging task to detect people in crowded scene owing to the heavy occlusion and small object characteristic of crowed people. In this paper we propose a multi-layer regional-based convolutional network to address the tasks of people detection in the crowded scenes. Unlike the traditional methods, we propose an end-to-end framework that uses the convolutional network as feature extractor followed by a proposal selecting network which generates candidate proposals online and performs the location optimization simultaneously. In the feature extracting section, we use unpooling layer and deconvolution layer to adjust the different layers' feature maps with different size and fusion the multi-layer feature maps as ROIs features. Then by separately detecting the heads of people, our system gets rid of the problems caused by occlusion and achieves robust detection results in the case that no matter whether the head is towards or backwards to the camera. Experimental results show that it greatly outperforms traditional methods.

Multi-Layer Proposal Network for People Counting in Crowded Scene