Crowd counting method combining density estimation and target detection

Abstract

The invention discloses a crowd counting method combining density estimation and target detection. The method comprises the following steps: S1, marking training pictures, training a density estimation network, a target detection network and a picture partition network by using the training pictures, and constructing a fusion network; s2, inputting the picture into a fusion network, and obtaining a sparse region and a dense region of the picture according to a picture partition network; s3, using a density estimation network in the dense region of the picture to obtain a head density map; using a target detection network in the sparse region of the picture to obtain a head bounding box density map; and fusing the two density maps through a fusion network to obtain a fused density map, and carrying out crowd counting. According to the method provided by the invention, the monitoring image is processed and analyzed in real time, so that the method can adapt to characteristics of various different scales, the influence of background factors on counting is reduced, the counting error is reduced, and the accuracy of crowd counting is improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a crowd counting method combining density estimation and target detection. Background technique [0002] In recent years, with the continuous expansion of the city scale, the urban population has continued to grow. Crowded crowds affect the normal traffic order, bring inconvenience to people's travel, and also have huge potential safety hazards. With the improvement of people's living standards in recent years, social activities and crowd gatherings in public places have become more common, and safety accidents have also shown a trend of high incidence. On the other hand, as people's demand for urban security management increases, the number of surveillance cameras in the city increases rapidly and their coverage becomes wider and wider. Numerous surveillance cameras constitute a huge surveillance network, which will generate massive monitoring data. Tradit...

AI Technical Summary

Problems solved by technology

At the same time, the background part outside the crowd may also affect the counting results

Therefore, the existing methods are difficult to meet the needs of complex scenarios

Images