Kaggle Amazon Contest
- 5. Co-occurence Matrixes Weather Labels Most Common Labels Rare Labels
- 7. Dataset JPG (RGB) and TIFF (GeoTIFF = RGB + near infrared) ~40k train set ~60k test set (40k + 20k)
- 8. First Model Keras 3 conv + 2 fc layers Adam (lr = 1e-3, 1e-4, 1e-5) sigmoid binary loss 0.909 on LB
- 9. Augmentation original 256x256 random crop 224x224 random rotate by 0, 90, 180 or 270 degree random horizontal or/and vertical flip
- 11. Finetuning
- 12. Ensemble 1) 5-layers CNN (0.925) 2) VGG-16 (0.927) 3) VGG-19 (0.928) 4) ResNet-50 x 3 (0.925-0.927) 5) ResNet-50 (sigmoid for non-weather + softmax for weather) (0.923) 6) ResNet-50 (weather) + ResNet-50 (non-weather) (0.925) 7) LightGBM on the ResNet-50 features (0.920) 4-threshold voting 0.93042 on Public, 0.92915 on Private
- 15. Panorama
- 16. Haze Removal Using Dark Channel Prior He, K., Sun, J., Tang, X.: Single Image Haze Removal Using Dark Channel Prior IEEE Trans. Pattern Anal. Mach. Intell. (2011)
- 17. Lessons Learned 1) Manually check your submits 2) Save experiments results 3) Try not to start too late 4) Optimize your code