Fatigue state detection method based on YOLOv5

The invention discloses a fatigue state detection method based on YOLOv5, and the method comprises the steps: firstly carrying out the enhancement of an original image through employing a Mosaic data enhancement method, then maintaining more complete sampling information for subsequent feature extraction through employing an attention mechanism, and finally carrying out the improvement of a loss function, thereby achieving the detection of a fatigue state. And the position relationship between the detection frame and the real frame can be better distinguished. Most of existing fatigue driving detection methods only adopt a single signal to judge the fatigue state of a driver, signal collection in the actual driving process is prone to being interfered by various factors, sometimes, collected data are not accurate enough, and due to the situations that the judgment basis is single, a trained model is poor in robustness, misinformation is prone to occurring and the like, the fatigue state of the driver is judge