This repository provides a user-friendly web interface for Nerfstudio using the Gradio library. The Nerfstudio Gradio WebUI allows users to easily train, visualize, process data, and export models without the need for complex command-line interactions.
- Training: Train Nerfstudio models directly from the web interface by selecting the dataset path and corresponding data parser.
- Visualization: Visualize trained models with Viser.
- Data Processing: Process training data by selecting the dataset path, output path, and processing method.
- Model Export: Export trained models by selecting the model configuration and output path.
-
Ensure that you have Nerfstudio (version >= 1.10) installed and set up properly.
-
Install the required dependencies by running the following command in the Nerfstudio environment:
pip install gradio
-
Clone this repository.
-
Start the Nerfstudio Gradio WebUI by running the following command (You may want to change the root_dir in webui.py):
python webui.py
-
Open a web browser and navigate to
http://localhost:7860to access the WebUI. -
Use the different tabs in the WebUI to perform various tasks:
-
Training: Select the dataset path and corresponding data parser, then click the "Train" button to start training. Monitor the training progress in the terminal.
-
Visualization: Choose the "Visualize" tab, select the configuration file for the trained model, and click the "Visualize" button to launch the Viser tool.
-
Data Processing: Choose the "Process Data" tab, select the dataset path, output path, and processing method. Click the "Submit" button to create a new folder if the output path doesn't exist. Then, click the "Process" button to start processing the data.
-
Model Export: Choose the "Export" tab, select the configuration file for the trained model and the output path. Click the "Export" button to start exporting the model.
-
-
Explore the different functionalities provided by the WebUI and enjoy a seamless experience with Nerfstudio!
Tested on Windows with 3070 and Linux with 2080ti. Multi-GPU untested. Subprocesses may not run properly on Windows when processing data.
This repo is still under developing and I'm currently an junior undergraduate student. Contributions to the Nerfstudio Gradio WebUI are super welcomed! If you encounter any issues or have suggestions for improvements, please open an issue or submit a pull request on the GitHub repository.
- Nerfstudio - The main repository for Nerfstudio.
- Gradio - The library used for building the web interface.