splat_one

splat_one is an integrated application that combines Gaussian Splatting and Structure from Motion (SfM) into a single workflow. It allows users to visualize and process image data through multiple interactive tabs, making it easier to generate 3D point clouds and analyze camera parameters.

Note: This project is currently under development. The Masks tab is not functioning properly at this time.

Tab Descriptions

• Images Tab
  Displays information about the imported images, such as file names, resolutions, and metadata.

• Masks Tab
  Provides functionality to generate masks for images.
  Currently under development; may not work as expected.

• Features Tab
  Shows the extracted image features, including keypoints and descriptors, for visualization and analysis.

• Matching Tab
  Displays the results of feature point matching, allowing you to inspect the quality and accuracy of the correspondences.

• Reconstruct Tab
  Visualizes the 3D point cloud reconstructed via SfM, along with camera positions and overall scene structure.

• Gsplat Tab
  Presents the output of the Gaussian Splatting process. You can adjust parameters to explore different renderings of the 3D point cloud.

Data Preparation

Before running the application, organize your images in the following directory structure inside the dataset folder:

splat_one
 └─ dataset
     └─ your_data
         └─ images

Place all the images you want to process under the images directory.

Docker Deployment

This application is designed to run via Docker. The Docker image is available as inuex35/splat_one. Please refer to the Dockerfile for more details on the installation.

To launch the Docker container with GPU support and proper X11 forwarding (for GUI display), run the following command:

docker run --gpus all \
    -e DISPLAY=host.docker.internal:0.0 \
    -v /tmp/.X11-unix:/tmp/.X11-unix \
    -v ${PWD}/dataset:/source/splat_one/dataset \
    -v C:\Users\$env:USERNAME\.cache:/home/user/.cache/ \
    -p 7007:7007 \
    --rm -it \
    --shm-size=12gb \
    inuex35/splat_one

Once inside the container, start the application by executing:

python main.py

Dependencies

• OpenSfM
• Gsplat

We use code from these repositories. Please note that the license for these components follows their respective original repositories.

Development Status

The project is under active development. Some features, such as the mask generation in the Masks tab, are not yet fully functional. Contributions via Issues and Pull Requests are welcome. License This project is licensed under the MIT License. See the LICENSE file for details.