Skip to content

Releases: MLMondays/mlmondays_data_imrecog

TAMUCC raw imagery (downsized, 3000-image subset)

02 Oct 04:25
@dbuscombe-usgs dbuscombe-usgs
0.1.1
cc35fd2
Compare
Choose a tag to compare
TAMUCC raw imagery (downsized, 3000-image subset) Latest
Latest

TAMUCC raw imagery (downsized, 2500-image subset)

5 individual zipped folders that each contain 500 images. Each image is 2808 x 1872 x 3 pixels, which is 50% of the original horizontal size

Imagery are in zipped 500-image batches:

tamucc_subset_1_500.zip
tamucc_subset_501_1000.zip
tamucc_subset_1001_1500.zip
tamucc_subset_1501_2000.zip
tamucc_subset_2001_2500.zip

Label data:

tamucc_subset.csv

Assets 8
Loading

ML Mondays image recognition datasets v1

22 Sep 18:31
@dbuscombe-usgs dbuscombe-usgs
0.1.0
016b723
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
ML Mondays image recognition datasets v1

Image recognition datasets for part 1 of ML Mondays

https://dbuscombe-usgs.github.io/MLMONDAYS/

https://github.com/dbuscombe-usgs/MLMONDAYS

TAMUCC

Publicly available imagery of coastal environments in Texas, provided by the Harte Research Institute at TAMUCC (Texas A&M University - Corpus Christi), funded by the Texas General Land Office (GLO).

Images such as the example above are oblique, taken from a low-altitude aircraft with approximate positions. In total, there are over 10,000 images covering the whole Texas coastline, each categorized by one of the following dominant classes ("developed" classes are ***):

* ***Exposed walls and other structures made of concrete, wood, or metal***
* Scarps and steep slopes in clay
* Wave-cut clay platforms
* Fine-grained sand beaches
* Scarps and steep slopes in sand
* Coarse-grained sand beaches
* Mixed sand and gravel (shell) beaches
* Gravel (shell) beaches
* ***Exposed riprap structures***
* Exposed tidal flats
* ***Sheltered solid man-made structures, such as bulkheads and docks***
* ***Sheltered riprap structures***
* Sheltered scarps
* Sheltered tidal flats
* Salt- and brackish-water marshes
* Fresh-water marshes (herbaceous vegetation)
* Fresh-water swamps (woody vegetation)
* Mangroves

The dataset is described further here. The dataset will be provided during the course, and is also available on Google Cloud Storage bucket gs://aju-demos-coastline-images/coastline

Acknowledgements

This dataset is courtesy of Texas A&M University (See https://storage.googleapis.com/tamucc_coastline/GooglePermissionForImages_20170119.pdf for details). Philippe Tissot, Associate Director, Conrad Blucher Institute for Surveying and Science, Texas A&M University - Corpus Christi; James Gibeaut, Endowed Chair for Coastal and Marine Geospatial Sciences, Harte Research Institute, Texas A&M University - Corpus Christi; and Valliappa Lakshmanan, Tech Lead, Google Big Data and Machine Learning Professional Services

Files

tamucc_full_2class.zip

tamucc_full_4class.zip

tamucc_subset_2class.zip

tamucc_subset_3class.zip

tamucc_subset_4class.zip

NWPU

publicly available benchmark for REmote Sensing Image Scene Classification (RESISC), created by Northwestern Polytechnical University (NWPU). The entire dataset contains 31,500 high-resolution images from Google Earth imagery, in 45 scene classes with 700 images in each class. The majority of those classes are urban/anthropogenic. We chose to use a subset of 11 classes corresponding to natural landforms and land cover, namely:

  • beach,
  • chaparral,
  • desert,
  • forest,
  • island,
  • lake,
  • meadow,
  • mountain,
  • river,
  • sea ice, and
  • wetland.

For more details, see this paper: https://www.mdpi.com/2076-3263/8/7/244

Files

nwpu.zip

Assets 8
Loading