Number-of-people estimation method based on deep learning semantic image segmentation

Abstract

The invention discloses a number-of-people estimation method based on deep learning semantic image segmentation. The number-of-people estimation method comprises a step 1 of constructing a training sample set including an original picture and a corresponding mask label picture; a step 2 of selecting or constructing a deep network model based on semantic image segmentation; a step 3 of training thesample set and obtaining a human head position prediction network model; and a step 4 of inputting a picture to be detected into the human head position prediction network model to obtain a mask picture, and obtaining the estimated number of people in the picture to be detected and position information of each person in the picture to be detected according to positions of points and the number ofthe points in the mask picture. Compared with an estimation method which is mostly employed in the prior art and based on image partitioning, the number-of-people estimation method is advantaged in that statistical errors brought by image partitioning can be avoided; the estimated number of people in the image area and the position of each pedestrian in the image area can be provided at the sametime.

Description

technical field [0001] The invention relates to the fields of image processing and computer vision, in particular to a method for estimating the number of people based on deep learning image semantic segmentation. Background technique [0002] Population estimation has application value in many scenarios. For example, in terms of public safety, too many crowds are prone to stampede and other accidents. When the crowd density reaches a certain scale, the number of people is controlled. For example, in urban or commercial planning, the area of ​​interest is analyzed The flow of people in the area can be used to efficiently plan commercial layouts; of course, if the location of each person in the area can be determined, more detailed crowd density information can be used. [0003] There are many crowd density or population estimation methods in the prior art, such as the Chinese invention patent application with the patent number CN201610536882, which is called a crowd density ...

AI Technical Summary

Problems solved by technology

[0003] There are many crowd density or population estimation methods in the prior art, such as the Chinese invention patent application with the patent number CN201610536882, which is called a crowd density estimation method based on convolutional neural network, which is based on the convolutional neural network of mixed-Pooling Network model, the final result is the density level (medium, high, low) of a certain area, but cannot give the estimated number of pedestrians in the area; another example is the Chinese invention patent application with the patent number CN201210434490, named as a A cross-camera adaptive crowd density estimation method, which cannot give an estimate of the number of people in the area

The above method cannot give the estimated number of people in the area, but there are also some patented methods that can give the estimated number of people, but most of them need to first divide the image into blocks, and then perform processing such as feature extraction. For example, the patent number is: CN201510336483 (a depth-based Intensive population estimation method based on learning), CN201610065279 (a crowd density estimation method based on integer programming), CN201610374700 (a crowd density estimation method based on multi-feature regression inheritance learning) Chinese invention patent applications, these methods are due to image Block processing is carried out, so that the head of the edge between the pixel block and the pixel block is segmented after block, which brings a lot of error and causes the estimation result to be inaccurate

Moreover, none of the above-mentioned existing technologies can solve the technical problem of how to give the distribution position of each pedestrian in the area

Images