1 paper · 1 filter
Simone Bonechi, Paolo Andreini, Monica Bianchini +1
Providing pixel-level supervisions for scene text segmentation is inherently difficult and costly, so that only few small datasets are available for this task. To face the scarcity…