VGTS: Visually Guided Text Spotting for Novel Categories in Historical Manuscripts

arXiv:2304.00746v4 Announce Type: replace
Abstract: In the field of historical manuscript research, scholars frequently encounter novel symbols in ancient texts, investing considerable effort in their identification and documentation. Although existing object detection methods achieve impressive performance on known categories, they struggle to recognize novel symbols without retraining. To address this limitation, we propose a Visually Guided Text Spotting (VGTS) approach that accurately spots novel characters using just one annotated support sample. The core of VGTS is a spatial alignment module …

abstract ancient texts arxiv cs.ai cs.cv detection detection methods documentation identification investing novel object performance research retraining scholars struggle text type