Formula reference number not captured, and incorrect formula recognition.

[luculli]

🔎 Search before asking | 提交之前请先搜索

• I have searched the MinerU Readme and found no similar bug report.
• I have searched the MinerU Issues and found no similar bug report.
• I have searched the MinerU Discussions and found no similar bug report.

🤖 Consult the online AI assistant for assistance | 在线 AI 助手咨询

• I have consulted the online AI assistant but was unable to obtain a solution to the issue.

Description of the bug | 错误描述

The formula reference number is not being captured (likely due to OCR limitations) and sometimes the recognized formula is incorrect. As an example, at the end of the formula, it should be "<v,w>", but it’s currently recognized as "<w,w>".

Despite these issues, it’s an amazing application. Thank you so much!

How to reproduce the bug | 如何复现

use the attached image

Operating System Mode | 操作系统类型

No response

Operating System Version| 操作系统版本

ubuntu 22.04

Python version | Python 版本

No response

Software version | 软件版本 (mineru --version)

No response

Device mode | 设备模式

No response

[dosubot]

Hi @luculli! I'm Dosu and I’m helping the MinerU team.

Your issue with formula reference numbers not being captured and formulas like <v,w> being misrecognized as <w,w> is a known challenge with OCR and formula recognition, especially for mathematical notation. Recent MinerU updates have improved accuracy by refining OCR detection thresholds, increasing batch sizes, and adding logic to remove low-confidence overlapping blocks, which helps reduce errors in formula and reference number extraction (PR #3139, PR #3240, PR #3275).

To get the best results, make sure you’re using the latest MinerU version (2.1.10 or newer), as it includes these fixes. You can also try tuning OCR parameters like det_db_box_thresh and enable_merge_det_boxes in your configuration, and ensure batch sizes for OCR and formula recognition are set to at least 16. If you’re processing PDFs with complex layouts, enabling image orientation and table classification can help improve recognition (PR #3242).

Some formula reference numbers and ends may still be missed due to inherent OCR/model limitations. For critical cases, manual review or post-processing of the output is recommended. If you need more targeted troubleshooting, sharing your MinerU version, config, and a sample file can help pinpoint further improvements.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}