SlugTrails dataset enables better indoor localization in large buildings
SlugTrails: An Egocentric Benchmark for Floor Plan Localization in Large Buildings
Computer Vision and Pattern Recognition
Summary
Finding your exact position inside large buildings using visual information and floor plans is difficult, especially when previous methods mostly worked in small homes. The authors created SlugTrails, a new collection of video and floor plan data from multiple floors of campus buildings, with precise position labels. They tested existing localization methods and found that training them on this new data improves accuracy a lot. This shows that better indoor localization depends on having large, realistic datasets like SlugTrails more than on changing model designs.
What this means in practice
- •For indoor navigation developers: Build accurate positioning systems for large public buildings using SlugTrails data as a benchmark with realistic egocentric video and floor plans.
- •For augmented reality app creators: Use fine-tuned models on SlugTrails to improve floor-plan-based visual localization for AR headsets in complex multi-floor environments.
Authors
Yunqian Cheng, Roberto Manduchi
Abstract
Floor-plan-based indoor visual localization enables infrastructure-free positioning, but most methods are developed and evaluated in small residential environments unlike the large public buildings of real deployment. We introduce SlugTrails, a floor plan localization benchmark for large indoor spaces under realistic egocentric sensing: $30$ Hz Aria glasses recordings across three campus buildings and six floors ($22089$ m$^2$ of floor plan outline), CAD-derived floor plans with semantic classes and circulation space masks, and trajectories aligned into the floor plan frame using laser-surveyed anchors. One protocol covers three practical ways of gathering geometry under a limited field of view -- a single walking frame, a stationary multi-view sweep, and a walking stream with odometry -- so methods designed for different regimes are compared on the same buildings and ground truth. Evaluating five representative geometric and learned systems under their native sensing configurations, we find that stock checkpoints (official released weights) are near zero on SlugTrails (at most $0.004$ R@1m30$^{\circ}$ on walking single frames), while fine-tuning on SlugTrails improves every trainable family on all three tasks (e.g., F$^3$Loc $0.0 \rightarrow 0.141$ single-frame and $0.03 \rightarrow 0.66$ sequential), with gains compounding as observations accumulate. The same fine-tuned weights also improve cross-dataset generalization on LaMAR with no LaMAR training (sequential R@1m $0.048 \rightarrow 0.143$ for F$^3$Loc and $0.063 \rightarrow 0.127$ for UnLoc), whereas train-from-scratch on SlugTrails alone stays far below fine-tuning from stock weights -- evidence that floor plan localization is currently limited by indoor data rather than by architecture. We release the dataset, protocols, and tools at https://github.com/Head-inthe-Cloud/SlugTrails.