gsplat is an open-source library designed for training and developing Gaussian Splatting methods. It features a front-end with Python bindings compatible with the PyTorch library and a back-end with highly optimized CUDA kernels. gsplat offers numerous features that enhance the optimization of Gaussian Splatting models, including improvements in speed, memory, and convergence times. Experimental results demonstrate that gsplat achieves up to 10% less training time and 4x less memory usage than the original implementation. Utilized in several research projects, gsplat is actively maintained on GitHub and welcomes contributions from the open-source community.
Gaussian splatting
N/A
N/A
Training time, memory usage
Open-source library
No
Yes
Optimized Gaussian splatting, PyTorch integration
No
GPU for training
Linux, Windows, macOS
Compatible with PyTorch
None
None
None
Yes
Active open-source community
Nerfstudio project team
N/A
N/A
N/A
None
None
Limited to Gaussian splatting methods
Machine learning research, computer graphics
Gaussian splatting model development, research projects
Machine learning researchers, developers
Integrates with PyTorch
N/A
Community support
None
Command-line interface
No
English
Open-source
Yes
Open-source community
None
None
1.0
Open-source library
No
Open-source
0.00
N/A
Apache License 2.0
01/01/2023
01/10/2023
+1-800-555-0199
Enhances Gaussian splatting model optimization
Yes