Reproduction implementation of VAAL(Variational Adversarial Active Learning) Segmantation Task. The classification task has an author implementation here
- Linux or macOS
- Python 3.8
- CPU compatible but NVIDIA GPU + CUDA CuDNN is highly recommended.
Please edit --data_dir in auguments.py to where your Cityscapes data is.
The code can simply be run using
python3 main.py
If you want to use GPU
python3 main.py --cuda
When using the model with different datasets or different variants, the main hyperparameters to tune are
--adversary_param --beta --num_vae_steps and --num_adv_steps
The results will be saved in results/accuracies.log.