Pattern Recognition Lab Member wins Third Place in ICDAR Competition

We are pleased to share that Fei Wu of the Computer Vision group has won third place in the “ICDAR 2025 Competition on Few-Shot Text Line Segmentation of Ancient Handwritten Documents”, where participants faced the difficult challenge of accurately segmenting text lines across three distinct ancient manuscripts using only three training samples per document. Fei Wu proposed a hybrid solution to this task: he additionally employed a full-image-sized instance segmentation model to improve the component connectivity of patch predictions from Wu et al.’s previously developed HookFormer semantic segmentation network. Please see the competition report for more details. He will receive the award at the 2025 International Conference on Document Analysis and Recognition, Wuhan, China.