Paper · ≈ 1 min read
MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction
Outgoing relations
- evaluated_on → Replica (Benchmark)
- evaluated_on → ScanNet (Benchmark)
- part_of → 3D/4D Vision and Reconstruction (ResearchField)
- uses_metric → LPIPS (Metric)
- uses_metric → PSNR (Metric)
- uses_metric → SSIM (Metric)
Connected node types
- Metric: 3
- Benchmark: 2
- ResearchField: 1
Mentions in the corpus
- part_of Evidence: State-of-the-art neural implicit methods allow for high-quality reconst…
- part_of Evidence: We demonstrate that depth and normal cues, predicted by general-purpose…
- part_of Evidence: We observe that geometric monocular priors improve performance both for…
- part_of Evidence: The monocular cues improve every backbone, with the hash-grid variant p…
- part_of Evidence: On ScanNet, Replica, and Tanks-and-Temples the method substantially imp…
- summarizes Project Pulse
- summarizes Field Overview — 3D/4D Vision and Reconstruction
Cross-references in raw data
Source provenance
examples/demo-corpus/data/research/papers/arxiv-2206-00665/abstract.md