Dense crowd detection method and device, electronic equipment and storage medium

Abstract

The invention provides a dense crowd detection method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining a to-be-detected image with a dense crowd; inputting the to-be-detected image into a trained target pedestrian detection model to obtain the to-be-detected image marked with a plurality of first prediction pedestrian frames; for at least two first prediction pedestrian frames corresponding to any area in the to-be-detected image, performing de-duplication processing on pedestrians in the at least two first prediction pedestrian frames to obtain the to-be-detected image marked with at least one second prediction pedestrian frame; and counting the target number of the second prediction pedestrian frames in the to-be-detected image, and determining the target number as the number of pedestrians in the to-be-detected image. In this way, the number of people in a dense place can be accurately detected.

Description

technical field [0001] The present disclosure relates to the technical field of target detection, and in particular, to a dense crowd detection method, device, electronic device, and storage medium. Background technique [0002] Object detection, also called object extraction, is an image segmentation based on the geometric and statistical features of objects, which combines the segmentation and recognition of objects into one, and its accuracy and real-time performance are an important capability of the entire system. Especially in the dynamic real-time tracking and positioning of the target, it has a wide range of application value in intelligent transportation system, intelligent monitoring system, military target detection and so on. [0003] Usually, when performing target detection on a crowd in a certain place, Region-Convolutional Neural Networks (R-CNN) is mainly used for target detection. Good recognition results, however, when performing person detection in dense...

AI Technical Summary

Problems solved by technology

[0003] Usually, when performing target detection on a crowd of people in a certain place, the Region-Convolutional Neural Networks (R-CNN) is mainly used for target detection, which can be obtained for scenes with fewer people or scattered people. Very good recognition results. However, when detecting people in dense places, it is difficult to identify a single target person from overlapping people when there are small targets, dense crowds, object occlusions, etc. in the panoramic image. lead to low recognition accuracy

Images