https://github.com/52north/landsat-classification
Science Score: 26.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (9.3%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: 52North
- Language: Python
- Default Branch: main
- Size: 231 MB
Statistics
- Stars: 0
- Watchers: 12
- Forks: 0
- Open Issues: 1
- Releases: 0
Metadata Files
README.md
ARCHIVED
This project is no longer maintained and will not receive any further updates. If you plan to continue using it, please be aware that future security issues will not be addressed.
TB17-landsat ML model for land cover classification
An Instance Segmentation model to classify different landcover classes using raw satellite imagery.
Datasets
Input: of Landsat 8 images Level 2 collection 2 using https://earthexplorer.usgs.gov/ web interface. \ The Multi-spectral Image consists of Blue, Green, Red, NIR, SWIR 1 and SWIR 2 corresponding to bands numbers (2, 3, 4, 5, 6, 7), respectively.
Label: Landcover 2015 from National Forest Information System \ The following classes were included in label data: - nochange - water - snowice - rockrubble - exposedbarren_land - bryoids - shrubland - wetland - wetlandtreed - herbs - coniferous - broadleaf - mixedwood
Preprocessing
The preparation of the train data consists of extracting pairs of input und output of the train and label data. This requires the datasets to be projected in the same spatial reference. Therefore, the landsat images were reprojected to match the same spatial reference of landcover dataset. After Datasets-registration patches with fixed size were extracted to prepare the train and label data.
Training
The model has u-net architecture consisting of 5 convolution and deconvolution layers. The model is trained to classify 4 different classes (water, herbs, coniferous and other) using the dice coefficient to evaluate accuracy. The model has reached total accuracy of 89% after learning for 120 epochs.
Testing or using the model
After the model loads the weights it can estimate raw bands images of landsat 8 using model.estimate_raw_landsat(path) as demonstrated in test.py. \
The raw landsat bands should be in one folder named as their originial Landsat Product Identifier L2 followed by the SR_B
The result classified_landcover.tiff is saved as a geo-referenced one-band GeoTiff in the same folder.
Owner
- Name: 52°North Spatial Information Research GmbH
- Login: 52North
- Kind: organization
- Email: info@52north.org
- Location: Münster
- Website: https://52north.org/
- Twitter: fivetwon
- Repositories: 261
- Profile: https://github.com/52North
Advancing spatial information infrastructures to foster open science
GitHub Events
Total
Last Year
Issues and Pull Requests
Last synced: over 1 year ago
All Time
- Total issues: 0
- Total pull requests: 1
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Total issue authors: 0
- Total pull request authors: 1
- Average comments per issue: 0
- Average comments per pull request: 0.0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Issue authors: 0
- Pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Top Authors
Issue Authors
Pull Request Authors
- EHJ-52n (1)
Top Labels
Issue Labels
Pull Request Labels
Dependencies
- Pillow *
- matplotlib *
- numpy *
- opencv-python *
- rasterio *
- requests *
- scikit-image *
- scikit-learn *
- tensorflow *