This repo contains a jupyter notebook with python code for individual fish ID modeling via CNN.
The dataset contains 6 images from each of 27 individual brook trout (162 images). Annotations include 2 labels: "IndID" = individual ID code (n=27) and "PopID" = source population code (n=3).
Photos and annotation files can be downloaded here: https://drive.google.com/drive/folders/1ye5yLqB9zVCRKy5svQOiX0Zs1FJQjDqh?usp=sharing
Original code was developed by Mona Arami and Natalya Rapstine. Than Hitt provided edits.
For background information on the data source: https://esip.figshare.com/articles/Deep_learning_identifies_individual_variation_and_population_structure_in_native_brook_trout/12602114/1
For background information on the project: https://www.usgs.gov/centers/lsc/science/enabling-ai-citizen-science-fish-biology