baselab/controlled-diet-methane-dataset

Controlled Diet Dataset for Methane Plume Detection

📊 Dataset Overview

This repository contains the CD dataset for methane plume detection, presented in the paper "Optical gas imaging and deep learning for quantifying enteric methane emissions from rumen fermentation in vitro". This dataset contains 4,885 thermal infrared images of methane plumes captured using FLIR GF77 optical gas imaging technology for semantic segmentation applications in agricultural emissions monitoring.

The images document methane emissions from controlled rumen fermentation experiments under three dietary conditions: Control diet (166-171 ppm), Low forage diet (300-334 ppm), and High forage diet (457-510 ppm).

📄 Associated Publication

"Optical gas imaging and deep learning for quantifying enteric methane emissions from rumen fermentation in vitro"
Mohamed G. Embaby, Toqi Tahamid Sarker, Amer AbuGhazaleh, Khaled R. Ahmed
IET Image Processing, January 19, 2025
DOI: 10.1049/ipr2.13327

📊 Dataset Details

Dataset Statistics

Class	GC Range (ppm)	Diet Description	Train	Val	Test	Total
Control	166-171	50:50 F:C ratio	1,079	138	133	1,350
Low Forage	300-334	20:80 F:C ratio	1,268	162	157	1,587
High Forage	457-510	80:20 F:C ratio	1,558	196	194	1,948

Class Mapping

• Class 0: Background
• Class 1: Control diet methane plumes (166-171 ppm, 50:50 F:C ratio)
• Class 2: Low forage diet methane plumes (300-334 ppm, 20:80 F:C ratio)
• Class 3: High forage diet methane plumes (457-510 ppm, 80:20 F:C ratio)

Mask Pixel Values:

⚠️ Important: There is a discrepancy between pixel values and class IDs in the mask files:

• Pixel value 0: Background (Class 0)
• Pixel value 1: Control diet (Class 1)
• Pixel value 2: High forage diet (Class 3) ⚠️
• Pixel value 3: Low forage diet (Class 2) ⚠️

📋 License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

📚 Additional Resources

For preprocessing methods and code examples, please refer to the GitHub repository.

🙏 Acknowledgments

This work was supported by the USDA National Institute of Food and Agriculture (NIFA) under Grant No. 2022-70001-37404.

📜 Citation

This work is published in IET Image Processing (2025, DOI: 10.1049/ipr2.13327).

@article{embaby2025optical,
  title={Optical gas imaging and deep learning for quantifying enteric methane emissions from rumen fermentation in vitro},
  author={Embaby, Mohamed G and Sarker, Toqi Tahamid and AbuGhazaleh, Amer and Ahmed, Khaled R},
  journal={IET Image Processing},
  year={2025},
  doi={10.1049/ipr2.13327}
}