Fatigue state detection method and device, electronic equipment and storage medium

Abstract

The invention relates to a fatigue state detection method. The method includes the steps of inputting a human eye sample and a mouth sample after segmenting the same face image; extracting first texture characteristics of the human eye sample and extracting second texture characteristics of the mouth sample; using a pre-trained strong classifier to detect the first texture characteristics and thesecond texture characteristics at the same time to obtain a human eye state and a mouth state; according to the human eye state and the mouth state, determining the percentage value PERCLOS value of the total time of eye closing to eye blinking of a user to which the face image belongs within a preset time, eye blinking frequency and the number of opening and closing of the mouth; determining whether or not the PERCLOS value is greater than a first preset value, whether or not the eye blinking frequency is greater than a second preset value, and whether or not the mouth opening number is greater than a third preset value; if any two or three of the three situations are true, determining that the user is in a fatigue state. The method can recognize the state of the eyes and the state of themouth at the same time, improve the efficiency of detecting the fatigue state, and at the same time, improve the recognition accuracy.

Description

technical field

[0001] The invention relates to the technical field of face detection, in particular to a fatigue state detection method, device, electronic equipment and storage medium. Background technique

[0002] With the acceleration of people's pace of life, many drivers often drive vehicles in a state of fatigue, which can easily lead to traffic accidents, so it is very important to give early warning to fatigue driving.

[0003] At present, the driver's fatigue state is identified by detecting the driver's eye blink and mouth state. Usually, in blink detection and mouth state detection, the positioning of eyes and mouth is mostly based on prior knowledge, and morphological image processing is performed according to the positions of eyes and mouth; or, grayscale projection based on template matching is used and geometric features. The former of these two types of methods is limited to using a large amount of prior knowledge, and the accuracy rate is not high, while ...

Images