Skip to content
/ CaC Public

Code for our paper 'Contrast and Clustering: Learning Nearest Pair Representations for Source-free Domain Adaptation'

Notifications You must be signed in to change notification settings

yukilulu/CaC

Repository files navigation

Code (based on pytorch 1.3, cuda 10.0, please check the 'requirements.txt' for reproducing the results) for our paper 'Contrast and Clustering: Learning Nearest Pair Representations for Source-free Domain Adaptation'(https://arxiv.org/pdf/2301.13428.pdf).

image

Our code is publicly available now!

1.(update in 2023/5/27): (1)fix the error in training file;(2)upload the file "A simple example of tuple ”match“.ipynb" to helo better understand the tuple "match"

Step1: Dataset preparing

Download the VisDA, Office-Home, and Office-31

Step2: Environment preparing

check the 'requirements.txt' for reproducing the results

Step3: Training

Go to 'Visda' folder and get the training code

Accuracy

image image image

Visualization

image image

About

Code for our paper 'Contrast and Clustering: Learning Nearest Pair Representations for Source-free Domain Adaptation'

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published