Semi-supervised data automatic labeling method based on deep learning, computer equipment and storage medium

The invention provides a semi-supervised data automatic labeling method based on deep learning, computer equipment and a storage medium. The method comprises the steps of obtaining a to-be-labeled multi-modal data set with aligned frames; manually annotating partial data in the to-be-annotated data set, and dividing the to-be-annotated data set into an annotated data set and an unannotated data set; automatically labeling the unlabeled data set based on a BEV deep learning algorithm to obtain a labeled data set and a new unlabeled data set; judging whether the new unlabeled data set is empty or not; and if the new unlabeled data set is not empty, evaluating the new labeled data set and the new unlabeled data set based on a dichotomy deep learning algorithm. According to the method designed by the invention, the data set is automatically labeled in a semi-supervised manner, and meanwhile, the labeled result is evaluated and optimized by adopting the dichotomy deep learning algorithm, so that the precision of t