Crowd monitoring method based on generative adversarial network, apparatus and device, and medium

Abstract

The invention provides a crowd monitoring method based on a generative adversarial network, which can be applied to the technical field of artificial intelligence. The method comprises the following steps: training a generative adversarial network, and after training is finished, estimating the number of people in a target monitoring area by using a generator model in the generative adversarial network to obtain a second crowd density estimation graph based on a second crowd image of the target monitoring area, wherein in the training process, a first crowd image is used as input of a generator model, and the generator model is trained to output a first crowd density estimation graph; meanwhile, taking the first crowd density truth value graph and the first crowd density estimation graph as input of a discriminator model so that the discriminator model can discriminate the similarity between the first crowd density truth value graph and the first crowd density estimation graph, and repeating the training process continuously until the discriminated similarity meets a preset threshold condition. The invention further provides a crowd monitoring device and equipment based on the generative adversarial network, a storage medium and a program product.

Description

technical field [0001] The present disclosure relates to the technical field of artificial intelligence, and more specifically relates to a crowd monitoring method, device, electronic device, computer-readable storage medium and program product based on a generative confrontation network. Background technique [0002] When crowds gather beyond a certain limit, it will inevitably lead to accidents, especially during the epidemic period, it is necessary to avoid gatherings. Therefore, it is very necessary to monitor the crowd cluster situation. [0003] In related technologies, a target detection method may be used to monitor the number of people. For example, the number of detected objects can be counted to obtain a final counting result by detecting and locating human body features (for example, human faces) in an image or video through camera monitoring. However, on the one hand, this method has a low accuracy rate and many false detections and missed detections in the sc...

AI Technical Summary

Problems solved by technology

However, on the one hand, this method has a low accuracy rate and many false detections and missed detections in the scene where the background and illumination changes are complex, the mutual occlusion of the crowd is serious, and the crowd scale changes greatly; on the other hand, due to the To detect human body features, there is a risk of personal body feature information being stolen

Images