WildGaussians: 3D Gaussian Splatting in the Wild
This addresses the problem of robust 3D reconstruction for applications in robotics, AR/VR, and computer vision by improving performance on in-the-wild data, though it is incremental as it builds on existing 3DGS methods.
The paper tackles the challenge of 3D scene reconstruction in uncontrolled, real-world environments with occlusions and varying illumination by introducing WildGaussians, which adapts 3D Gaussian Splatting to handle such conditions and achieves state-of-the-art results while maintaining real-time rendering speeds.
While the field of 3D scene reconstruction is dominated by NeRFs due to their photorealistic quality, 3D Gaussian Splatting (3DGS) has recently emerged, offering similar quality with real-time rendering speeds. However, both methods primarily excel with well-controlled 3D scenes, while in-the-wild data - characterized by occlusions, dynamic objects, and varying illumination - remains challenging. NeRFs can adapt to such conditions easily through per-image embedding vectors, but 3DGS struggles due to its explicit representation and lack of shared parameters. To address this, we introduce WildGaussians, a novel approach to handle occlusions and appearance changes with 3DGS. By leveraging robust DINO features and integrating an appearance modeling module within 3DGS, our method achieves state-of-the-art results. We demonstrate that WildGaussians matches the real-time rendering speed of 3DGS while surpassing both 3DGS and NeRF baselines in handling in-the-wild data, all within a simple architectural framework.