search

Gorilla-Lab-SCUT / SRDC-CVPR2020

Code release for Unsupervised Domain Adaptation via Structurally Regularized Deep Clustering (CVPR2020-Oral).
MIT License
80 stars 16 forks source link

can't reproduce the result in paper with default opts #8

Open zhao-ht opened 3 years ago

zhao-ht commented 3 years ago

I failed to reproduce the experimental results in your paper. For example, on the A-W dataset of office31, the best result is only 92.70. I used the default hyperparameters, and the only thing I changed was the path of the target dataset: python main.py --data_path_source ./data/datasets/Office31/ --src amazon --data_path_target ./data/datasets/Office31/- -tar webcam --data_path_target_t ./data/datasets/Office31/ --tar_t webcam I want to know the hyperparameters used in the results in your paper. This is very important for me to understand your research. Thank you!

ljjcoder commented 3 years ago

@zhao-ht sorry to bother you. I am a stranger in UDA. I want to ask, how to get webcam_half file. In the link given by the author, I did not find *_half folder.