Strong generalization multi-type license plate recognition system based on grouping feature decoupling

A strong generalization multi-type license plate recognition system based on grouping feature decoupling comprises a feature extraction module comprising a backbone network, a group module, an indirect supervision module and a feature decoupling decoding module comprising a character decoder. Extracting license plate type information from the instance features as type noise through a two-stage processing mode of extraction and decoupling of fusion grouping features including implicit multi-dimensional visual representation and explicit license plate type information, and then removing the type noise from the instance features through grouping feature decoupling; and finally, predicting the license plate content by adopting a character decoder based on a single character attention mechanism. According to the method, the training result on the non-real data obtained through synthesis can be well generalized to a real scene, the cost is greatly reduced, and meanwhile, the performance is ensured. 一种基于分组特征解耦的强泛化性多