Skip to content
forked from swamiviv/LSD-seg

Learning from Synthetic Data: Addressing Domain Shift for Semantic Segmentation

Notifications You must be signed in to change notification settings

vcg-uvic/LSD-seg

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

README

Code for Semantic Segmentation for Unsupervised Domain Adaptation

Prerequisites:

Install pytorch (Version 0.2) and torchvision
Install fcn (pip install fcn)
Install OpenCV (pip install opencv-python)

Datasets:

We will need two datasets for our experiments - SYNTHIA and CITYSCAPES. Please download the datasets into data folder from the following links

Please download SYNTHIA-RAND-CITYSCAPES subset of the SYNTHIA dataset. SYNTHIA: http://synthia-dataset.net/download-2/

CITYSCAPES: https://www.cityscapes-dataset.com/

To run the code, go to code folder and run the following command:

python run_script.py

This assumes that the data is downloaded and paths are set accordingly. Options can be modified directly in the train.py script.

Please change the dataroot path, and logdir path accordingly. This will run the code and save the models in logdir folder.

To evaluate the trained model, run

python eval_cityscapes.py --dataroot [] --model\_file [] -- method []

About

Learning from Synthetic Data: Addressing Domain Shift for Semantic Segmentation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.9%
  • Shell 0.1%