Research map / Detection and segmentation / Open-set and few-shot detection
Joint Color-Spatial-Directional Clustering and Region Merging (JCSD-RM) for Unsupervised RGB-D Image Segmentation
Abul Hasnat, Olivier Alata, Alain Trémeau
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 region merging after JCSD clustering (ablation) | JCSD · NYUD2 | GTRC: 0.46 -> 0.6 | Improved |
| add region merging after JCSD clustering (ablation) | JCSD · NYUD2 | VoI: 2.68 -> 2.2 | Improved |
| use independent color/depth/normal model instead of single GMM over all features (ablation) | JCSD · NYUD2 | GTRC: 0.43 -> 0.46 | Improved |
| use independent color/depth/normal model instead of PCA-reduced GMM (ablation) | JCSD · NYUD2 | GTRC: 0.34 -> 0.46 | Improved |
| use directional distribution for normals instead of Gaussian (ablation) | JCSD · NYUD2 | GTRC: 0.41 -> 0.46 | Improved |
| use independent model instead of All_GMM (ablation) | JCSD · NYUD2 | computation time (sec): 45 -> 31 | Improved |
| use region merging with independent model instead of All_GMM + RM (ablation) | JCSD-RM · NYUD2 | VoI: 2.41 -> 2.21 | Improved |
| use JCSD-RM instead of UCM-RGBD | segmentation · NYUD2 | BDE: 9.11 -> 8.97 | Improved |
| use JCSD-RM instead of UCM-RGBD on boundary F-measure | segmentation · NYUD2 | BFM: 0.63 -> 0.61 | Hurt |
| use JCSD-RM instead of UCM-RGBD for speed | segmentation · NYUD2 | time (sec): 110 -> 38 | Improved |
| vary number of clusters k from 15 to 25 (ablation) | JCSD-RM · NYUD2 | BDE: 9.4 -> 8.97 -> 8.99 | No clear effect |
| lower th_b from 0.2 to 0.1 (ablation) | JCSD-RM · NYUD2 | VoI: 2.20 -> 2.34 | Hurt |
Closest papers by meaning
- Locality-Sensitive Deconvolution Networks With Gated Fusion for RGB-D Indoor Semantic Segmentation (CVPR 2017 · similarity 0.85)
- RDFNet: RGB-D Multi-Level Residual Feature Fusion for Indoor Semantic Segmentation (ICCV 2017 · similarity 0.85)
- PDDM: Pseudo Depth Diffusion Model for RGB-PD Semantic Segmentation Based in Complex Indoor Scenes (AAAI 2025 · similarity 0.84)
- UCTNet: Uncertainty-Aware Cross-Modal Transformer Network for Indoor RGB-D Semantic Segmentation (ECCV 2022 · similarity 0.84)
- 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans (CVPR 2019 · similarity 0.84)
- UnScene3D: Unsupervised 3D Instance Segmentation for Indoor Scenes (CVPR 2024 · similarity 0.83)
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.
