SplitFusion: Simultaneous Tracking and Mapping for Non-Rigid Scenes

TL;DR

SplitFusion is a new RGB-D SLAM framework that separates scenes into rigid and non-rigid parts for improved tracking and reconstruction, effectively capturing dynamic elements like moving humans.

Contribution

It introduces a deep learning-based scene segmentation method combined with separate rigid and non-rigid tracking and reconstruction techniques.

Findings

Accurately maps environments with dynamic non-rigid objects

Effectively tracks both rigid and non-rigid scene components

Reconstructs moving targets such as humans with high fidelity

Abstract

We present SplitFusion, a novel dense RGB-D SLAM framework that simultaneously performs tracking and dense reconstruction for both rigid and non-rigid components of the scene. SplitFusion first adopts deep learning based semantic instant segmentation technique to split the scene into rigid or non-rigid surfaces. The split surfaces are independently tracked via rigid or non-rigid ICP and reconstructed through incremental depth map fusion. Experimental results show that the proposed approach can provide not only accurate environment maps but also well-reconstructed non-rigid targets, e.g. the moving humans.