You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If you use this code or the attached files for research purposes, please cite
@inproceedings{robbiano2021adversarial,
title = {Adversarial Branch Architecture Search for Unsupervised Domain Adaptation},
author = {Robbiano, Luca and Ur Rahman, Muhammad Rameez and Galasso, Fabio and Caputo, Barbara and Carlucci, Fabio Maria},
year = 2022,
booktitle = {2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
volume = {},
number = {},
pages = {1008--1018},
doi = {10.1109/WACV51458.2022.00108}
}
Software requirements
CUDA
Python 3.6 or newer
PyTorch 1.6 or newer
Other Python libraries listed in requirements.txt
Hardware requirements
10 GB available on each GPU
Optional but strongly recommended: a cluster capable of running at least 8 parallel GPU jobs
Run experiments
To launch an ABAS run (OfficeHome, source Art, target Clipart):
The script launch_slurm_stub.sh needs to be customized according to your cluster setup. A similar script can be developed for other schedulers, like PBS.
Once the job is done, a result.pkl file will be produced. To analyze the results, run