SemanticFusion: Dense 3D Semantic Mapping with Convolutional Neural Networks

TL;DR

SemanticFusion combines CNN-based semantic predictions with dense SLAM to produce real-time, detailed 3D semantic maps, improving accuracy over single-frame methods for indoor RGB-D video.

Contribution

It introduces a system that fuses CNN semantic predictions with dense SLAM, enabling real-time 3D semantic mapping with improved accuracy.

Findings

Fusing multiple views improves semantic labeling accuracy.

System operates at approximately 25Hz for real-time use.

Enhanced 2D semantic segmentation performance over single-frame predictions.

Abstract

Ever more robust, accurate and detailed mapping using visual sensing has proven to be an enabling factor for mobile robots across a wide variety of applications. For the next level of robot intelligence and intuitive user interaction, maps need extend beyond geometry and appearence - they need to contain semantics. We address this challenge by combining Convolutional Neural Networks (CNNs) and a state of the art dense Simultaneous Localisation and Mapping (SLAM) system, ElasticFusion, which provides long-term dense correspondence between frames of indoor RGB-D video even during loopy scanning trajectories. These correspondences allow the CNN's semantic predictions from multiple view points to be probabilistically fused into a map. This not only produces a useful semantic 3D map, but we also show on the NYUv2 dataset that fusing multiple predictions leads to an improvement even in the 2D semantic labelling over baseline single frame predictions. We also show that for a smaller reconstruction dataset with larger variation in prediction viewpoint, the improvement over single frame segmentation increases. Our system is efficient enough to allow real-time interactive use at frame-rates of approximately 25Hz.