search

ioos / seafloor-sampling-ml

Seafloor imagery machine learning project, GSoC 2021
MIT License
1 stars 2 forks source link

Habitat Classification Model to detect Sea Floor #8

Closed lohithmunakala closed 3 years ago

lohithmunakala commented 3 years ago

Adding code that has a notebook to train a model to classify whether the image of the transect is pure SAND/MUD or not. This PR addresses issue #7.

A model classifying only pure SAND/MUD was finalized as when biologists classify an image, they segment each shell/gravel based on 50 points on the image. As all the transects have more than 50% of pure MUD/SAND images, it was decided to train a model only on pure MUD/SAND images to reduce the manual time taken for each of the transects that are covered.

UPDATE : This notebook has to be trained on other data, aka the GoPro image data, which is better in terms of quality and blurriness.

review-notebook-app[bot] commented 3 years ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB