All-St3R SLAM: Real-time RGB Dynamic SLAM with Pixel-wise Motion Masking

This research proposes All-St3R SLAM, a real-time RGB dynamic SLAM framework that precisely removes moving objects at the pixel level without predefined labels or segmentation models. The system combines ego-motion prediction from DUSt3R with point tracking from All-Tracker, achieving 10 FPS performance on standard 30 FPS video streams, which significantly improves upon state-of-the-art dense dynamic SLAM approaches (0.5-3 FPS).