Consensus Entropy | 共识熵

English | 中文

English

A Python library for calculating consensus entropy between multiple strings, particularly useful for OCR result analysis. Uses Levenshtein distance to calculate string differences.

This library is the official implementation of our paper: Consensus Entropy: Harnessing Multi-VLM Agreement for Self-Verifying and Self-Improving OCR

Citation

If you use this library in your research, please cite our paper:

@misc{zhang2025consensusentropyharnessingmultivlm,
      title={Consensus Entropy: Harnessing Multi-VLM Agreement for Self-Verifying and Self-Improving OCR}, 
      author={Yulong Zhang and Tianyi Liang and Xinyue Huang and Erfei Cui and Xu Guo and Pei Chu and Chenhui Li and Ru Zhang and Wenhai Wang and Gongshen Liu},
      year={2025},
      eprint={2504.11101},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2504.11101}
}

Installation

pip install consensus-entropy

Usage

Basic Usage

from consensus_entropy import calculate_consensus_entropy

# Calculate consensus entropy for multiple OCR results
ocr_results = [
    "Hello World",
    "Hello Wrld",
    "Hallo World"
]

# Calculate entropy values for each result
entropy_values = calculate_consensus_entropy(ocr_results, task_type="ocr")
print(entropy_values)  # [0.1667, 0.3333, 0.3333]

Get Best OCR Result

from consensus_entropy import get_best_ocr_result

# Get the OCR result with lowest entropy
ocr_results = ["Test1", "Test2", "Text2"]
best_result, best_entropy = get_best_ocr_result(ocr_results, task_type="ocr")
print(f"Best result: {best_result}")
print(f"Entropy: {best_entropy:.4f}")

Features

• Calculate normalized Levenshtein distance
• Compute consensus entropy for multiple strings
• Get the best OCR result with lowest entropy
• Support for both English and Chinese text
• Type hints for better IDE support
• Optimized for OCR tasks

Requirements

• Python 3.7+
• numpy
• python-Levenshtein

Notes

• Currently only supports OCR task type
• Input string list must contain at least two elements
• All inputs will be converted to string type

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

中文

一个用于计算多个字符串之间共识熵的Python库，特别适用于OCR结果分析。使用Levenshtein距离来计算字符串之间的差异。

本库是我们论文的官方实现：Consensus Entropy: Harnessing Multi-VLM Agreement for Self-Verifying and Self-Improving OCR

引用

如果您在研究中使用了本库，请引用我们的论文：

@misc{zhang2025consensusentropyharnessingmultivlm,
      title={Consensus Entropy: Harnessing Multi-VLM Agreement for Self-Verifying and Self-Improving OCR}, 
      author={Yulong Zhang and Tianyi Liang and Xinyue Huang and Erfei Cui and Xu Guo and Pei Chu and Chenhui Li and Ru Zhang and Wenhai Wang and Gongshen Liu},
      year={2025},
      eprint={2504.11101},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2504.11101}
}

安装

pip install consensus-entropy

使用方法

基本用法

from consensus_entropy import calculate_consensus_entropy

# 计算多个OCR结果的共识熵
ocr_results = [
    "人工智能",
    "人工智障",
    "人工智能",
    "人工智惠"
]

# 计算每个结果的熵值
entropy_values = calculate_consensus_entropy(ocr_results, task_type="ocr")
print(entropy_values)  # [0.1667, 0.2500, 0.1667, 0.2500]

获取最佳OCR结果

from consensus_entropy import get_best_ocr_result

# 获取熵值最低的OCR结果
ocr_results = ["测试文本1", "测试文本2", "文本2"]
best_result, best_entropy = get_best_ocr_result(ocr_results, task_type="ocr")
print(f"最佳结果: {best_result}")
print(f"熵值: {best_entropy:.4f}")

功能特点

• 计算标准化Levenshtein距离
• 计算多个OCR结果的共识熵
• 获取熵值最低的最佳OCR结果
• 支持中文和英文文本
• 类型提示支持
• 针对OCR任务优化的算法

系统要求

• Python 3.7+
• numpy
• python-Levenshtein

注意事项

• 目前仅支持OCR任务类型
• 输入字符串列表至少需要两个元素
• 所有输入都会被转换为字符串类型处理

许可证

本项目采用MIT许可证 - 详见LICENSE文件

贡献

欢迎提交Pull Request来改进这个项目。