1 paper
Xinyan Zu, Haiyang Yu, Bin Li +1
Text segmentation is a challenging vision task with many downstream applications. Current text segmentation methods require pixel-level annotations, which are expensive in the cost…