Abnormal behavior detection method and device and storage medium

Abstract

The invention discloses an abnormal behavior detection method, an abnormal behavior detection device and a storage medium, gives full play to the advantages of high response and low delay of an edge server on the basis of using deep learning, and provides an abnormal behavior detection method based on non-data knowledge transmission. According to the method, the synthesized image data set for training can be provided for the edge model on the premise of not accessing the original data set and not using any natural image and mark data, and the diversity of the synthesized image is improved by utilizing the divergence between the cloud model and the edge model. Experiments prove that the method is high in feasibility, the trained model can operate on edge equipment, the accuracy rate is high during abnormal behavior detection, the method has the advantages of low delay and fast response, and the defect that a huge and complex cloud model cannot operate on the edge equipment is effectively overcome.

Description

technical field [0001] The present invention relates to abnormal behavior detection technology, in particular to an abnormal behavior detection method, device and storage medium. Background technique [0002] With the continuous development of computer intelligence technology, the abnormal behavior detection of crowds is becoming more and more common in the fields of intelligent security, smart city and so on. Most of the current abnormal behavior detection uses cloud service technology, but in practical applications, it often needs to calculate huge data and get instant feedback, which will occupy a large amount of network bandwidth and bring great pressure to the communication network. [0003] Edge computing is a supplement and extension of cloud computing. It disperses large-scale services to edge devices for processing, making data processing more timely and transmission more secure. Models deployed on the cloud have high requirements for hardware computing power and m...

AI Technical Summary

Problems solved by technology

However, the images synthesized by these methods are often not realistic, and are quite different from natural images, so they cannot be used for knowledge distillation

[0004] Additionally, generative adversarial networks (GANs) can be used to generate high-fidelity synthetic images [8] , this method trains two opposing networks (generator and discriminator) to generate synthetic images, although the resolution of the synthetic images generated by this method has increased, the generator for training GAN still needs access to the original data

Images