Method, device, and computer device for judging driving behavior category

Abstract

The invention relates to the field of image recognition, in particular to a method, a device and a computer device for judging a driving behavior category. The method comprises the following steps: obtaining an image to be recognized; Extracting a feature map of the image to be recognized by a feature extraction network based on a convolution neural network; Analyzing the feature map by using a full convolution neural network to obtain a driving behavior class of the driver; Outputting the driving behavior class. The invention uses the convolution nerve depth feature extraction network to automatically extract the deep-level feature of the image data, overcomes the shortcomings of the prior art that the artificial prior knowledge is required to extract the feature process of the image datawhich is too complex and inefficient, and the feature extraction is simpler and more efficient, and the feature extraction representation is higher at the same time. Full convolution neural network is used to predict the coordinates of human joints and facial feature points simultaneously as the input features of the classifier, which makes the classification of driving behavior more accurate.

Description

technical field [0001] The present invention relates to the field of image recognition, in particular to a method, device, calculation and equipment for judging the type of driving behavior. Background technique [0002] Road transport traffic safety is the focus of social attention. The unsafe driving behavior of drivers during driving is the main factor affecting road traffic safety. Real-time monitoring, analysis, identification and early warning of drivers' driving behavior during driving are important An effective method to ensure road transport traffic safety. [0003] The existing technologies all use visual algorithms to identify and analyze the driver's driving behavior, and then give early warning to unsafe driving behavior. However, the existing technologies generally compare the manually defined target feature templates with the monitored target features. The artificially defined feature templates cannot fully reflect the essential characteristics of the data, a...

AI Technical Summary

Problems solved by technology

However, the existing technologies generally compare the manually defined target feature templates with the monitored target features. The artificially defined feature templates cannot fully reflect the essential characteristics of the data, and the generalization ability is low, resulting in low accuracy. The correct process requires a lot of calculations, which reduces the efficiency of behavior recognition

[0004] It can be seen that the existing technology has the problems of large amount of calculation and low accuracy in the identification of driver's unsafe driving behavior, which needs to be improved.

Images