Paper Overview
- Field: Computer Vision (CV)
- Authors: Yejun Zhang, Xinjue Wang, Zihan Wang, Esa Rahtu, Juho Kannala
- Published: 2026-07-02
- arXiv: 2607.02486
- Category: cs.CV
- Underutilize local geometry cues
- Lack global context among keypoints
- Overfit to a single keypoint detector
- Sets a new state of the art among descriptor-free methods
- Reduces the 75th-percentile rotation error by 89%
- Reduces translation error by up to 90%
- Zero-shot generalizes to unseen keypoint detectors
- Narrows the gap with descriptor-based pipelines
Summary
Descriptor-free visual localization eliminates high-dimensional descriptor storage, preserves scene privacy, and simplifies map maintenance, yet its accuracy still lags far behind descriptor-based pipelines. The authors attribute this gap to insufficient geometric discriminability in geometry-only matching.
Without visual appearance, current methods:
The authors further observe that descriptor-free matching naturally enables multi-detector training, as heterogeneous keypoints can be optimized in a shared geometry-only space without aligning descriptor spaces.
GeoMix
Based on these insights, GeoMix is a descriptor-free 2D-3D matching framework that strengthens geometric discriminability at three levels:
1. Local level: Orientation-aware and distance-aware embeddings enrich neighborhood aggregation with fine-grained spatial structure. 2. Global level: Learnable context nodes aggregate and redistribute scene-level information via cross-attention, resolving ambiguities beyond local receptive fields. 3. Training level: Mix-Training leverages this detector-agnostic geometric space to learn representations across multiple keypoint detectors.
Results
Extensive experiments on MegaDepth, Cambridge Landmarks, 7Scenes, and Aachen Day-Night show that GeoMix:
*Auto-collected on 2026-07-06*