ImprovedVBGS: Real-time Continual Variational Bayes Gaussian Splatting
Computer Vision and Pattern RecognitionRobotics
Summary
The authors focus on improving a method called Variational Bayes Gaussian Splatting (VBGS) used for continuously reconstructing 3D scenes in applications like robotics. The original VBGS was too slow for real-time use because it processed all data points every time a new frame came in. They developed ImprovedVBGS, which speeds things up a lot by only focusing on spatially relevant data and avoiding unnecessary calculations. Their method runs about 1680 times faster while still keeping good reconstruction quality, making it more practical for real-time tasks.
Authors
Damani Mguni-Coker
Abstract
On-the-fly reconstruction is a key requirement for many applications in robotics and autonomous navigation. Variational Bayes Gaussian Splatting (VBGS) enables continual learning without replay buffers using Coordinate Ascent Variational Inference (CAVI), but its per-frame iterations over all observed points make it too slow for real-time use with strict memory and latency requirements. We present ImprovedVBGS, an accelerated framework for on-the-fly continual reconstruction. This is achieved primarily through (i) spatially truncated variational inference, and (ii) improved reassignment that uses forwarding, truncation and eliminates wasteful dynamic recompilation. On the NeRF synthetic dataset, we reduce mean per-frame latency from ~84.0 s to ~0.050 s on an RTX 3070 Ti, a 1680x speed-up while maintaining reconstruction quality.