Published September 2, 2026
| Version v1
Model
Open
Handwritten Document Dataset Splits
Authors/Creators
-
1.
Luleå University of Technology
Description
You can download the Hi-SAM and YOLO models' weights trained on the mixed dataset for our paper, “Generalization of Text Line Segmentation for HTR in Historical Documents,” from the URL below. Since Hi-SAM training is computationally expensive, you can use our pretrained weights to detect text lines in unseen documents.