nam-trinh / ucc-ai-quest-baseline

Install dependencies

conda create -n ucc python=3.10
pip install torch torchvision lightning albumentations matplotlib
conda activate ucc

Unzip warmup.zip in data folder so it has the following structure

data/warmup/img/train
data/warmup/img/valid
data/warmup/ann/train
data/warmup/ann/valid

Training

python train.py

Prepare results for submission

After training, the checkpoints are stored in folder checkpoints. We need to prepare a file named "results.json" for submission on CodaLab. Use the script prepare_results_for_submission.py and replace the checkpoint path

model = SegModel.load_from_checkpoint("checkpoints/epoch=6-val_loss=0.47-val_high_vegetation_IoU=63.55-val_mIoU=66.48.ckpt")

and the phase (warmup/public/private) and the split accordingly

ds = SegDataset(phase="warmup", split="valid", transform=ToTensorV2())

then run`

python prepare_results_for_submission.py

there should be a file "results.json" generated in the current directory. Zip the file

zip results.zip results.json

Done! You should be able to submit the file results.zip now, good luck!