Optical imaging and machine learning to identify and enumerate early developmental stages of fish species

This upload contains the following: 1) Color training library2) Grayscale training library3) Color enumeration runs4) Grayscale enumeration runs 5) Larval Fish Paper Code6) Larval Fish Paper installation7) HowToInstallAndRun8) HowToBuild The color training library consists of the 42 classes used in this paper. Each .bmp image in each of the folders have been validated as part of that class. The grayscale training library consists of the 5 classes used in experiment 6 of this paper. Each .bmp image in each folder has been validated as part of that class. The validated images for both color and grayscale training libraries provide a reproducible source to test and retest various classification algorithms and image decomposition approaches. The color enumeration runs zipped folder contains a series of subfolders that contain 10 enumeration runs for each species and stage of development. Inside each enumeration run is a list of folders, which represent predictions made by the SAO. Inside each prediction folder are subfolders that contain validated images. These enumeration runs used the full color training library and the CNN algorithm. The grayscale enumeration runs zipped folder also contains a series of subfolders that contain 10 enumeration runs for each species. The first series of subfolders represent the predictions made by the grayscale SAO, while the next series of subfolders contain the validated images. The larval fish code and larval fish paper installation contains the necessary codes and instructions for reproducibility. The HowToInstallAndRun and HowToBuild files provide the information necessary to utilize these files.