A traffic anomaly event detection method based on instance-level attention mechanism

Abstract

The present invention provides a method for detecting abnormal traffic events based on an instance-level attention mechanism, using a self-learning multi-target detector as a front-end detection module to complete vehicle detection in the scene, and then through the VGG19 network to detect video frames and target objects Perform feature extraction, and input the features into the attention module to assign corresponding attention weights to the detected vehicles in the scene, and finally complete the detection of abnormal events through LSTM and Softmax. The beneficial effects of the present invention are: using self-learnable multi-target detector as the front-end detection module, can self-learn, and gradually improve detection performance; add attention mechanism, improve the accuracy of event detection; can classify abnormal events, make The monitoring side can accurately make corresponding decisions according to the event type.

Description

technical field [0001] The invention relates to the field of pattern recognition, in particular to a method for detecting abnormal traffic events based on an instance-level attention mechanism. Background technique [0002] In recent years, with the rapid development of social economy and the acceleration of urbanization, the number of people's cars has increased significantly. While cars have brought great convenience to people's daily travel, they have also brought traffic management to traffic management. Congestion, frequent traffic accidents and other serious challenges, traffic anomalies caused by the increase of urban traffic vehicles are increasing year by year. Traffic abnormal events refer to events in traffic scenes that are different from normal driving, such as illegal parking, road collisions, and traffic jams. Nowadays, intelligent traffic video surveillance is an important part of urban traffic management. Surveillance cameras are deployed on major arterial ...

AI Technical Summary

Problems solved by technology

[0004] Based on the problems existing in the above-mentioned background, the present invention proposes a traffic anomaly event detection method based on an instance-level attention mechanism. This method uses an algorithm in deep learning to solve the problem that traditional methods require a large amount of manual processing of massive data, and proposes a method An attention mechanism enables the model to focus on abnormal areas and improve the detection accuracy of events

Images