Abnormal driving behavior detection method based on multi-task depth convolution neural network

Abstract

The invention discloses an abnormal driving behavior detection method based on a multi-task depth convolution neural network, which is divided into two phases: the training phase of multi-task depth learning face detection model and the phase of abnormal face behavior detection, face abnormal behavior detection phase is based on the multi-task depth-learning face detection model generated in the training phase of multi-task depth-learning face detection model to detect driving abnormal behavior, in the training phase of multi-task depth-learning face detection model, multi-task depth-learningface detection model is trained by multi-task convolution neural network framework, and then face detection loss function, face orientation loss function, other feature point loss function and total loss function are calculated according to the multi-task depth-learning face detection model. The method applies the multi-task depth-learning face detection model to extract features of face, face orientation and other features of face, and then uses the extracted features to detect abnormal driving behaviors, which can detect abnormal driving behaviors quickly and accurately.

Description

technical field [0001] The invention relates to abnormal driving behavior detection in the field of intelligent transportation, in particular to a method for detecting abnormal driving behavior based on a multi-task deep convolutional neural network. Background technique [0002] According to the statistics of WHO in 2009, 1.23 million people die from traffic accidents every year worldwide. But we know that in the Korean War, the number of deaths in the entire war was almost one million. In other words, the number of people who die in traffic accidents every year is almost equal to the number of deaths in a very tragic war. According to WHO statistics, there are as many as 1.23 million deaths caused by traffic accidents worldwide every year; and 90% of traffic accidents are caused by drivers, such as fatigue driving, inattention, speeding, and safety awareness. Weak and so on. [0003] At present, abnormal driving behavior detection is mainly divided into two categories: ...

AI Technical Summary

Problems solved by technology

[0007] Although the industry has proposed many good deep learning architectures for face detection, the sample size required for deep learning training is too large and difficult to train.

Images