FreeView: training-free distractor rejection enables robust 3D reconstruction from uncurated photos
zhenjun_zhao · x · 2026-10-30
FreeView is a training-free framework for robust 3D reconstruction from uncurated image collections. It uses collection-level consensus (DINOv2 Top-K similarity) to automatically select a representative reference view, then adaptively rejects distractors by combining appearance and geometric cues — rejecting only when both are low — integrating with VGGT and π³ without manual curation or calibration. Across Phototourism, ETH3D and On-the-Go benchmarks at varying distractor levels, it achieves high clean-view retention and distractor rejection, markedly improving pose and reconstruction accuracy over reference-conditioned baselines.
More from Research
- Self-Direction turns agent mistakes into LoRA lessons, lifting GAIA success from 38.8% to 55.8% — rohanpaul_ai · 2026-10-30
- Preprint: AlphaGenome pervasively underestimates causal variant effects, undermining fine-mapping — anshulkundaje · 2026-10-30
- One Researcher on 22 NeurIPS 2026 Papers: What Should Author Lists Mean? — qberthet · 2026-10-30
- Stanford's I3L boosts whole-body robot deployment success by 68.3 points — jiajunwu_cs · 2026-10-30
- Dev ditches unanimous-vote stopping rule for a judge in multi-AI debate project — lluisantoni · 2026-10-30
- Google's AI Search Uses Your Knowledge Graph Facts Without Linking, RESONEO Research Finds — dejanseo · 2026-10-30