Research map / 3D vision / 3D generation
Unsupervised Multi-View CNN for Salient View Selection of 3D Objects and Scenes
Ran Song, Wei Zhang, Yitian Zhao, Yonghuai Liu
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 view distinction (VD) and weighted sum pooling (WSP) instead of max pooling (ablation) | UMVCNN-30 · Best View Selection benchmark (68 objects) | mean VSE: 0.384 -> 0.367 | Improved |
| use 30 views instead of 20 for the multi-view representation (ablation) | UMVCNN-30 · Best View Selection benchmark (68 objects) | mean VSE: 0.367 -> 0.366 | No clear effect |
| set output dimension C to 10 instead of 30 (ablation) | UMVCNN · Best View Selection benchmark (68 objects) | mean VSE: 0.367 -> 0.379 | No clear effect |
| set output dimension C to 20 instead of 30 (ablation) | UMVCNN · Best View Selection benchmark (68 objects) | mean VSE: 0.367 -> 0.382 | No clear effect |
| set output dimension C to 40 instead of 30 (ablation) | UMVCNN · Best View Selection benchmark (68 objects) | mean VSE: 0.367 -> 0.380 | No clear effect |
| unsupervised UMVCNN-30 instead of weakly supervised Deep mesh distinction | salient view selection · Best View Selection benchmark (68 objects) | mean VSE: 0.380 -> 0.367 | Improved |
| unsupervised UMVCNN-30 instead of mesh saliency | salient view selection · Best View Selection benchmark (68 objects) | mean VSE: 0.430 -> 0.367 | Improved |
| unsupervised UMVCNN-30 instead of view area | salient view selection · Best View Selection benchmark (68 objects) | mean VSE: 0.517 -> 0.367 | Improved |
Closest papers by meaning
- Unsupervised Multi-View CNN for Salient View Selection and 3D Interest Point Detection (IJCV 2022 · similarity 0.88)
- Volumetric and Multi-View CNNs for Object Classification on 3D Data (CVPR 2016 · similarity 0.88)
- Toward Unsupervised, Multi-Object Discovery in Large-Scale Image Collections (ECCV 2020 · similarity 0.87)
- The 3D-PC: a benchmark for visual perspective taking in humans and machines (ICLR 2025 · similarity 0.85)
- MOVIS: Enhancing Multi-Object Novel View Synthesis for Indoor Scenes (CVPR 2025 · similarity 0.85)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · 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.
