1 paper
Wenjun Liu, Qian Wu, Yifeng Hu +1
We introduce SELECT (Scene tExt Label Errors deteCTion), a novel approach that leverages multi-modal training to detect label errors in real-world scene text datasets. Utilizing an…