Image classification using PyTorch on (almost) STL-10 dataset
Loading data from a locally stored folder "dataset" (added to .gitignore). Dataset is almost similar to STL-10: https://ai.stanford.edu/~acoates/stl10/
The difference is that images are resampled for each of 10 classes:
- local dataset has 1000 train and 300 test images per class;
- original STL-10 has 500 train and 800 test images per class.