[Model suggestion] DiT for document image classification/layout analysis/table detection

[lmmx]

The paper on DiT (Document Image Transformer) was revised last week with code and pre-trained models, and I think it’d be a great candidate for inclusion in the LayoutParser library’s model zoo.

• arXiv: DiT: Self-supervised Pre-training for Document Image Transformer
• GitHub repo
• Interactive demo: Document Layout Analysis with DiT
• Interactive demo: Document Image Classification with DiT

[lolipopshock]

Thank you @lmmx ! We are planning for the next versions and will take it into account :)