Deep Learning Based Facial Area Detection Method for Pedestrian Violations on Roads
Pedestrians may exhibit various postures,including walking,stationary,standing,or squatting,and there are also differences in their clothing and appearance among different pedestrians.In addition,pedestrian targets may also be affected by various factors such as surrounding environment, lighting conditions,and obstructions,making facial area detection more complex.To this end,a deep learning based facial area detection method for road pedestrians violating regulations is proposed.Com-bining composite residual learning methods and deep convolutional networks in deep learning,denoising is applied to facial images of pedestrians violating regulations on the road.Introduce the concept of double-layer subspace to extract facial features of road pedestrians violating regulations.Apply the YCrCb color space to determine the facial areas of pedestrians who violate regulations on the road.The experimental results show that the research method can accurately detect the facial areas of non compli-ant pedestrians in traffic monitoring,and the detection error divergence is low.
deep learningroadspedestrian violationsfacial areatest method