Refactor

Open nshaud opened this issue 5 years ago • 1 comments

[x] Rewrite disjoint sampling method
[x] Move sklearn models into their own file
[x] Rewrite the Dataset
[x] Add parallelization option (see #32)
[x] Use a unique IGNORED_INDEX value for all ignored pixels
[x] Use sklearn.metrics everywhere needed (especially in validation)
[x] Move data exploration/data visualization functions into their own file
[x] Rewrite the build_dataset function
[x] Main script uses a main function
[x] Unify val and test function
[ ] Deal with the varying tensor sizes when using : spectra (1D), images (2D), cubes (3D).
[x] Unify segmentation and classification datasets
[ ] Use Sequential API for simple models
[x] Add --overlap options for training and test
[ ] Use scheduler/auto LR reduction (see #22)
[ ] Save output image after training
[ ] Rewrite data augmentation as torchvision.transforms (see #33)
[ ] Move downloaders into their own script
[ ] Add other class balancing schemes (see #39)
[ ] Add IoU/dice score loss
[ ] Improve cross-validation support
[ ] Optional: Simplify dataset configuration
[ ] Optional: Support defining a dataset as a collection of HSI images and GT masks

Nov 27 '20 18:11 nshaud

Where is the output image after training saved,I really can't find it

Dec 06 '23 14:12 3097108366