method: TencentOCR V12023-03-21
Authors: Haoxi Li, Weida Chen, Huiwenshi, Mei Jiang, Jing Lv, Chunchao Guo, Hongfa Wang, Wei Liu
Affiliation: TencentOCR
Description: We use multiple DBNet and Cascade MaskRCNN integrated detection results built with various backbones, combined with the Parseq English recognition model for recognition, and then end-to-end tracking and trajectory recognition results based on the improved ByteTrack.