03_Directory Structure and Tests - Doodleverse/segmentation_gym GitHub Wiki
After Gym downloads, we recommend you make a directory structure that mirrors below, entirely separate from the src directory contains the Gym source code.
Trust us, put it somewhere that is entirely separate from the src directory contains the Gym source code.
A config folder will store *.json config files, which determine how the dataset is made and the model is trained. The the images and labels folders contain jpg format images and greyscale label images respectively. The weights folder holds *.h5 files - the weights and biases for your trained tensorflow model. The holdOutSet folder is for jpg format images and is optional. The weights, npzForModel, and modelOut directories should be left empty - they will be populated with files by running Gym programs, as explained below.
/Users/Someone/my_segmentation_gym_datasets
β βββ config
β | βββ *.json
β βββ capehatteras_data
| | βββ fromDoodler
| | | βββimages
| β | βββ *.jpg
β | | βββlabels
| β | βββ *.jpg
| | βββnpzForModel
| β βββ *.npz
β | βββtoPredict
| β βββ *.jpg
β βββ modelOut
β βββ *.png
β βββ weights
β βββ *.h5
After datasets have been made, model-ready data files in npz format are now populated in the npzForModel.
/Users/Someone/my_segmentation_Gym_datasets
β βββ config
β | βββ *.json
| βββimages
β | βββ *.jpg
β |ββlabels
β | βββ *.jpg
| |ββnpzForModel
β | βββ *.npz
β |ββholdOutSet
β | βββ *.jpg
β βββ modelOut
β βββ weights
After model training, the weights folder is populated with model files and modelOut contains example model predictions, model training curves, and other handy figures in png format.
/Users/Someone/my_segmentation_Gym_datasets
β βββ config
β | βββ *.json
| βββimages
β | βββ *.jpg
β |ββlabels
β | βββ *.jpg
| |ββnpzForModel
β | βββ *.npz
β |ββholdOutSet
β | βββ *.jpg
β βββ modelOut
β | βββ *.png
β βββ weights
β βββ *.h5