Dangerous driving behavior real-time detection method based on deep learning

Abstract

The invention proposes a dangerous driving behavior real-time detection method based on deep learning, and the method comprises the steps: building a front car image collection system, collecting a training sample, carrying out the manual marking, and building a dangerous driving behavior data set; proposing a spatial pyramid pooling convolution depth confidence normalized classification network (SPP-CDBRNet) model according to the characteristics of the dangerous driving behavior data set based on a deep learning method; carrying out the preprocessing of the built data set, carrying out the training of the SPP-CDBRNet model through a random gradient descent method with a momentum and the data set after preprocessing, and obtaining an SPP-CDBRNet which precisely recognize whether there is a dangerous driving behavior (behaviors of using a cellphone and smoking during driving); carrying out the detection of a front car image through the determined SPP-CDBRNet model, and achieving the real-time detection of the dangerous driving behavior. The method can effectively improve the detection precision of dangerous driving behaviors, is good in instantaneity and mobility, and is good in application prospect.

Description

technical field [0001] The invention relates to problems related to the detection of dangerous driving behaviors in the field of intelligent transportation, and in particular to a real-time detection method of dangerous driving behaviors based on deep learning. Background technique [0002] With the advancement of machinery manufacturing and vehicle engineering technology and the improvement of people's economic and living standards, the number of cars and drivers in our country is constantly increasing. The advancement of transportation tools has brought convenience to people's lives, but it has also caused frequent traffic accidents. , has caused a great threat to the safety of people's lives and property. Among them, illegal driving behaviors such as answering the phone and smoking are the main causes of accidents. This has also attracted the attention of the government, universities and other research institutions and automobile companies. The detection, reminder, and su...

AI Technical Summary

Problems solved by technology

This type of method requires the driver to wear some physiological and psychological detection devices, which are complicated to operate and inconvenient to wear. They are easily affected by the driver's individual factors and cannot truly and effectively reflect the actual situation. Moreover, due to the influence of price and wearing comfort, it is not easy to popularize. ;(2) Detection system based on vehicle sensor detection, such as Zhang Xibo. Driver fatigue state detection method based on steering wheel operation [J]. Journal of Tsinghua University: Natural Science Edition, 2010, 50(7): 1072-1076

This type of method is to install various sensors that can detect the state of the vehicle, and analyze the driver's behavior through the state of the vehicle and make a forecast. This method requires high hardware and is expensive, and because the driving behavior of different drivers is quite different, Very easy to cause interference, high false alarm rate

[0005] The method of non-contact detection is currently mainly by installing a camera in the car and performing detection and analysis through traditional image processing methods, such as application number 201510585266.1, and the title of the invention is "A Driver Applicable to Answering and Calling in Multiple Attitudes" Behavior Detection Method", this method is less disturbed than the contact detection method, easy to use, and cheap, but it is difficult to meet the real-time detection requirements and is easily affected by light and the driver's appearance, and image analysis needs to be based on However, the traditional image processing method loses global information when extracting local texture features in the first step, resulting in a decline in accuracy.

Images