Skip to content

ppingzhang/2DGS_ImageCompression

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This is the code for "2D Gaussian Splatting for Image Compression".

Training command:

python main.py --primary_samples=100 --backup_samples=100 --num_embeddings=32 --outf=100_32 --num_epochs=2000 --image_dir=./data/kodim01.png

Argument Possible values
--primary_samples the number of primary samples
--backup_samples the number of backup samples
--num_embeddings the number of embeddings
--outf the output directory
--num_epochs the number of epochs for training
--image_dir the directory of the image

Alternatively, you can follow the settings in the train_all.py file.

During training, the best model will be saved.

Test command:

python main.py --eval=True --primary_samples=100 --backup_samples=100 --num_embeddings=32 --outf=100_32 --num_epochs=2000 --image_dir=./data/kodim01.png

PS:

  • I'll be very happy if you could accelerate it or convert it into CUDA language. Try it!
  • I've exported the environment.yml file from my Conda environment for reference purposes.
  • Please feel free to contact me if you have any problem.

About

2D Gaussian splatting for image compression

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages