Research map / 3D vision / Face and shape
Puzzle Similarity: A Perceptually-guided Cross-Reference Metric for Artifact Detection in 3D Scene Reconstructions
Nicolai Hermann, Jorge Condor, Piotr Didyk
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 |
|---|---|---|---|
| use PuzzleSim cross-reference matching instead of CrossScore | artifact localization · Human artifact dataset (12 scenes, mean) | Pearson correlation: 0.510 -> 0.615 | Improved |
| use PuzzleSim instead of CrossScore | artifact localization · Human artifact dataset (12 scenes, mean) | Spearman correlation: 0.378 -> 0.474 | Improved |
| use PuzzleSim instead of CrossScore | artifact localization · Human artifact dataset (12 scenes, mean) | std of Pearson across scenes: 0.204 -> 0.120 | Improved |
| use PuzzleSim instead of no-reference PaQ-2-PiQ | artifact localization · Human artifact dataset (12 scenes, mean) | Pearson correlation: 0.402 -> 0.615 | Improved |
| use PuzzleSim instead of CrossScore | artifact localization · Scene stump | Pearson correlation: 0.585 -> 0.505 | Hurt |
| use PuzzleSim instead of CrossScore | artifact localization · Scene garden | Pearson correlation: 0.748 -> 0.675 | Hurt |
| use PuzzleSim instead of CrossScore | artifact localization · Scene treehill | Pearson correlation: 0.300 -> 0.717 | Improved |
| use SqueezeNet features instead of VGG (ablation) | PuzzleSim · Human artifact alignment | map quality | No clear effect |
| combine three network layers instead of one (ablation) | PuzzleSim · Human artifact alignment | map quality | Improved |
| use more than three downsamplings of the image resolution (ablation) | PuzzleSim · Human artifact alignment | map quality | No clear effect |
| compare patches across all reference positions instead of the same position (ablation) | PuzzleSim · camera shifts | robustness | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CrossScore — CrossScore: A Multi-View Approach to Image Evaluation and Scoring (ECCV 2024) · better on 11 of 14 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Puzzles: Unbounded Video-Depth Augmentation for Scalable End-to-End 3D Reconstruction (NeurIPS 2025 · similarity 0.87)
- SALART-VQA: Diagnosing Whether VLMs Understand Salient Artifacts in Generated Images (NeurIPS 2026 · similarity 0.86)
- Sparse3D: Distilling Multiview-Consistent Diffusion for Object Reconstruction from Sparse Views (AAAI 2024 · similarity 0.86)
- Re-assembling the past: The RePAIR dataset and benchmark for real world 2D and 3D puzzle solving (NeurIPS 2024 · similarity 0.85)
- Quantifying and Alleviating Co-Adaptation in Sparse-View 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.85)
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 2025 · similarity 0.85)
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.
