EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding

[📜 Paper] • [🤗 HF Models] • [🐱 GitHub]

Repo for "EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding"

2024.12.9: The paper has been accepted by AAAI-25.
2024.12.31: The EvoChart corpus （qa-pair, images, matadata） has been open-sourced.

This repository contains the code and data for the paper "EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding".

🔥🔥🔥 Hugging Face Model: huggingface.co

🚀🚀🚀 The EvoChart-QA Benchmark can be downloaded at: drive.google.com

📂📂📂 EvoChart corpus can be downloaded at: pan.quark.cn

EvoChart-QA Benchmark Answer Verification:

• If the is_clear is True, string matching should be used to determine correctness.
• If the is_clear is False, a 5% tolerance is allowed.

Citation

If you find it helpful, please kindly cite our paper.

@misc{huang2024evochartbenchmarkselftrainingapproach,
      title={EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding}, 
      author={Muye Huang and Lai Han and Xinyu Zhang and Wenjun Wu and Jie Ma and Lingling Zhang and Jun Liu},
      year={2024},
      eprint={2409.01577},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2409.01577}, 
}