Crowd Counting Based on Multiresolution Density Map and Parallel Dilated Convolution

Figure 2

Crowd counting network model diagram based on multiscale density map fusion dilated convolution. (bx means multiresolution density map; HD means final high-resolution density map; MDM is responsible for retaining spatial contact information; PDM aims to expand receptive fields and further extract features; Up-2 represents the upsampling with a step size of 2; Conv7-256 represents the convolution kernel of 7 × 7 × 256; Conv3 + Upsample means that there is a series of convolution and upsampling processes).