Environment

The environment for our project is similar to that of our main baseline Feature_3DGS.

conda create --name df_3dgs python=3.8
conda activate df_3dgs

pip install torch==2.0.1+cu118 torchvision==0.15.2+cu118 --index-url https://download.pytorch.org/whl/cu118

pip install -r requirements.txt

dataset structure

<location>
|---images
|   |---<image 0>
|   |---<image 1>
|   |---...
|---test_images
|   |---<image 0>
|   |---<image 10>
|   |---...
|---train_images
|   |---<image 5>
|   |---<image 15>
|   |---...
|---sparse
    |---0
        |---cameras.bin
        |---images.bin
        |---points3D.bin

LSeg encoder

Download the LSeg model demo_e200.ckpt from the Google drive and put it to lseg_encoder/checkpoint.

Run

For detailed steps, please refer to the comment.sh file.

Acknowledgement

Our repository is developed based on the excellent work of the following open-source projects:Feature_3DGS, LangSplat,gsplat,3D Gaussian Splatting. We would like to extend our sincere gratitude to the authors for making their codebases available to the public.