The aim of our computer vision final project is to construct an image classification model that can identify whether a given satellite image captures one of six types of methane-emitting facilities. You can read our paper here: https://tinyurl.com/2bt3dkw4.
Based on repo from Stanford ML Group: https://github.com/stanfordmlgroup/aicc-aut21-multi-task
To train the model, run:
python main.py train
To evaluate the model on the test set, run:
python main.py test --ckpt_path <checkpoint path>