Skip to content

Deep Learning based Remote Sensing Methods for Methane Detection in Airborne Hyperspectral Imagery

License

Notifications You must be signed in to change notification settings

satish1901/Methane-detection-from-hyperspectral-imagery

Repository files navigation

Methane-detection-from-hyperspectral-imagery

H-MRCNN introduces fast algorithms to analyze large-area hyper-spectral information and methods to autonomously represent and detect CH4 plumes. This repo contains 2 methods for processing different type of data, Single detector works on 4-channels data and Ensemble detectors works on 432-channels raw hyperspectral data recorded from AVIRIS-NG instrument.

Satish Kumar*, Carlos Torres*, Oytun Ulutan, Alana Ayasse, Dar Roberts, B S Manjunath.

Official repository of our WACV 2020 paper.

This repository includes:

  • Source code of single-detector and ensemble detectors(H-MRCNN) built on Mask-RCNN.
  • Training code for single-detector and ensemble detectors(H-MRCNN)
  • Pre-trained ms-coco weights of Mask-RCNN
  • Annotation generator to read-convert mask annotation into json.
  • Modified spectral library of python
  • Example of training on your own dataset

supported versions Library GitHub license

The whole repo folder structure follows the same style as written in the paper for easy reproducibility and easy to extend. If you use it in your research, please consider citing our paper (bibtex below)

Citing

If this work is useful to you, please consider citing our paper:

@inproceedings{kumar2020deep,
  title={Deep Remote Sensing Methods for Methane Detection in Overhead Hyperspectral Imagery},
  author={Kumar, Satish and Torres, Carlos and Ulutan, Oytun and Ayasse, Alana and Roberts, Dar and Manjunath, BS},
  booktitle={2020 IEEE Winter Conference on Applications of Computer Vision (WACV)},
  pages={1765--1774},
  year={2020},
  organization={IEEE}
}

Requirements

  • Linux or macOS with Python ≥ 3.6
  • Tensorflow <= 1.8
  • CUDA 9.0
  • cudNN (compatible to CUDA)

Installation

  1. Clone this repository
  2. Install dependencies
pip install -r requirements.txt

Single-detector

Running single-detector is quite simple. Follow the README.md in single_detector folder

single_detector/README.md

Ensemble-detector

For Running ensemble-detector we need some pre-processing. Follow the README.md in emsemble_detector folder

ensemble_detector/README.md

About

Deep Learning based Remote Sensing Methods for Methane Detection in Airborne Hyperspectral Imagery

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published