/Understanding-the-Amazon-Rainforest-from-Satalite-Images-with-Computer-Vision.

Multilabel classification problem using CNN with pytorch

Primary LanguagePython

Understanding-the-Amazon-Rainforest-from-Satalite-Images-with-Computer-Vision.

Computer Vision Kaggle challenge of multi-label classification of satelite images

J.Castillo, J.D.García, J.F.Suescún, Universidad de Los Andes, Bogotá, Colombia 2020.

Requierements:

-Pytorch: conda install pytorch torchvision cudatoolkit=10.2 -c pytorch

-Spectral: conda install -c conda-forge spectral

-Torch summary: pip install torchsummary

-TensorFlow: conda install -c conda-forge tensorflow

-Tensorboard_logger: pip install tensorboard_logger

3. Test

Run:

python -W ignore test.py --model saved-models/AmazonResNet101-run-3.pth.tar --output_file "pred_x.csv" --batch_size 8

For testing in Kaggle:

kaggle competitions submit -c planet-understanding-the-amazon-from-space -f pred_x.csv -m "Run-x"