Summary of Unitable: Towards a Unified Framework For Table Recognition Via Self-supervised Pretraining, by Shengyun Peng et al.
UniTable: Towards a Unified Framework for Table Recognition via Self-Supervised Pretraining
by ShengYun Peng, Aishwarya Chakravarthy, Seongmin Lee, Xiaojing Wang, Rajarajeswari Balasubramaniyan, Duen Horng Chau
First submitted to arxiv on: 7 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents UniTable, a training framework that unifies table recognition (TR) by combining pixel-level inputs with self-supervised pretraining from diverse unannotated tabular images. The framework trains on a unified task-agnostic objective: language modeling, and achieves state-of-the-art performance on four large TR datasets, surpassing existing methods and general vision-language models like GPT-4o, GPT-4-turbo with vision, and LLaVA. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Tables are hard for machines to understand because they use special rules that humans take for granted. The UniTable framework helps machines better understand tables by combining different approaches into one training method. This makes it good at three tasks: understanding table structure, reading cell content, and finding cell boundaries. The paper shows that UniTable does well on many big datasets and can even beat other methods designed specifically for tables. |
Keywords
* Artificial intelligence * Gpt * Pretraining * Self supervised
