Abnormity detection model generation method and device and abnormal event detection method and device

Abstract

The embodiment of the application discloses an abnormity detection model generation method and device. According to one specific embodiment, the method comprises the following steps: acquiring a plurality of sample image frame sequences; on the basis of a first image and a second image, training a prediction frame generator included in an initial model, the prediction frame generator comprising amulti-level feature extraction network and a generation network, and the feature extraction network being used for extracting feature information of different depths of the first image and fusing thefeature information; using the generation network for generating a prediction frame by utilizing the fused feature information; training a frame discriminator included in the initial model based on the prediction frame and the second image; and in response to completion of training, determining the trained initial model as an abnormity detection model. According to the embodiment of the invention,a method of fusing a plurality of feature information of different depths is adopted, so that the generated prediction frame is closer to reality, and the accuracy of anomaly detection is improved.

Description

technical field [0001] The embodiments of the present application relate to the field of computer technology, and in particular to a method and device for generating an anomaly detection model, and a method and device for detecting an abnormal event. Background technique [0002] Anomaly detection (Anomaly Detection) problem is a common application of machine learning algorithms. Let a system learn some normal features from many unlabeled data, so as to be able to diagnose abnormal data, we call this process anomaly detection. The so-called anomaly detection is to find objects that are different from most objects, in fact, it is to find outliers. Anomaly detection has different definitions in different fields. Anomaly detection in video refers to identifying events that do not match the expected behavior and distinguishing normal events from abnormal events. [0003] In current anomaly detection methods, feature reconstruction using normal training data is a common strateg...

AI Technical Summary

Problems solved by technology

When using the method of manually designing features, the accuracy of the results cannot be guaranteed because the dictionary has not been trained on abnormal events and is often incomplete.

There are also some problems with the use of deep learning-based methods. The capacity of deep neural networks is very high, and large reconstruction errors for abnormal events may not necessarily occur, resulting in inaccurate results for final anomaly detection.

Images