Driving environment identification method and device, and storage medium

The invention discloses a driving environment identification method, which comprises the steps of extracting features in an image in a driving environment through ResNet50, and classifying weather and illumination conditions through different classification heads of the ResNet50 to obtain an identification result; designing a label of a corresponding type according to the weather information data, training the deep learning model by using the label, and finding out a wrong label through repeated confidence learning; and S3, taking a union set from the error tags found through repeated confidence learning in the step S2, removing the union set, training the multi-head model again by using the remaining tags after the error tags are removed, and obtaining the accuracy of identifying the weather tags. According to the method, related data of weather and illumination in the driving environment can be obtained at the same time, and the weather and illumination conditions are obtained through one-time training. 本发明