A multimodal transformer that fuses language, image, and polygon geometry links recognized words into reading-order phrases on historical maps better than prior methods.
In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LIGHT: Multi-Modal Text Linking on Historical Maps
A multimodal transformer that fuses language, image, and polygon geometry links recognized words into reading-order phrases on historical maps better than prior methods.