Research map / 3D vision / Face and shape
Doppelgangers: Learning to Disambiguate Images of Similar Structures
Ruojin Cai, Joseph Tung, Qianqian Wang, Hadar Averbuch-Elor, Bharath Hariharan, Noah Snavely
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add keypoint and match masks as network input (ablation) | Doppelgangers classifier · Doppelgangers test | average precision: 64.7 -> 93.6 | Improved |
| use LoFTR masks instead of SIFT+RANSAC masks (ablation) | Doppelgangers classifier · Doppelgangers test | average precision: 87.6 -> 93.6 | Improved |
| align the input image pair with an affine transform (ablation) | Doppelgangers classifier · Doppelgangers test | average precision: 92.3 -> 93.6 | Improved |
| add flip data augmentation (ablation) | Doppelgangers classifier · Doppelgangers test | average precision: 93.6 -> 95.2 | Improved |
| replace CNN backbone with vision transformer (ablation) | Doppelgangers classifier · Doppelgangers test | average precision: 95.2 -> 95.5 | No clear effect |
| use learned spatial-distribution classifier instead of thresholding number of matches | visual disambiguation · Doppelgangers test (16 scenes) | average precision: 86.0 -> 95.2 | Improved |
| use DINO features for classification | visual disambiguation · Doppelgangers test (16 scenes) | average precision: 62.0 vs 95.2 | Hurt |
| filter COLMAP scene graph edges with the learned classifier | COLMAP SfM · SfM disambiguation (16 scenes) | number of scenes correct: 0 -> 12 | Improved |
| raise COLMAP match-count threshold to 150 as a simple disambiguation | COLMAP SfM · SfM disambiguation (16 scenes) | number of scenes correct: 0 -> 5 | Mixed |
| tune classifier threshold per scene on Cereal, Street and Church on Spilled Blood | Doppelgangers SfM filter · SfM disambiguation | reconstruction correctness | Improved |
Closest papers by meaning
- Doppelgangers++: Improved Visual Disambiguation with Geometric 3D Features (CVPR 2025 · similarity 0.92)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.85)
- MESC-3D:Mining Effective Semantic Cues for 3D Reconstruction from a Single Image (CVPR 2025 · similarity 0.85)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · similarity 0.84)
- MATCHA: Towards Matching Anything (CVPR 2025 · similarity 0.84)
- Bridging Viewpoint Gaps: Geometric Reasoning Boosts Semantic Correspondence (CVPR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
