CCAI_MASKGEN_ADVENT_TEMP

What needs to be done before run the code?

1. Create a new subfolder under the root named pretrained_models
2. Download the pretrained DeepLabv2 model and save it in the folder
3. Change the comet_ml work_space and user in the config files (shared/config.yml)

What needs to be done before test the code?

1. Copy the model that you want to test to the root
2. The model should be named like model_{30000}.pth. 30000 can be replaced to any int
3. Go to shared/config.yml find TEST-Model-test_iter and change the number to the number in the name of your .pth file

How to run the code?

$ pip install .
$ python train_CCAI.py # by default setting
$ python train_CCAI.py --cfg ./shared/advent.yml # by updated settings in advent.yml

How to test the code?

$ python test_CCAI.py

How to uninstall the package?

$ pip uninstall ADVENT

Acknowledge

ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation [https://github.com/valeoai/ADVENT]