ematvey / kaggle-carvana

Semantic segmentation with PyTorch

Carvana Semantic Segmentation - solution

[Challenge link]

This solution gets 0.9945 dice, which is within 0.25% from current top, but still places in the middle of the leaderboard (competition requires pixel-perfect segmentation).

This is mainly for-fun solution, without any special tricks (no ensembling, no data augmentation), with very little hyperparameter tuning.

Architecture: slightly customized UNet (so that it can process images in native 1918x1280 resolution)

Framework: PyTorch

Running

1. Extract competition data in input/
2. Preprocess data with python dataset.py (20 min)
3. Train to convergence with python train.py (12+ hours)
4. If necessary, generate submission with python make_submission.py

License

MIT