Research map / Graph learning / Knowledge and temporal graphs
Leveraging Commonsense for Object Localisation in Partial Scenes
Francesco Giuliari, Geri Skenderi, Marco Cristani, Alessio Del Bue, Yiming Wang
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 commonsense knowledge (MLP with ConceptNet via GCN) | MLP · 2D localisation | LSR: 0.143 -> 0.163 | Improved |
| add concept nodes with learned embeddings (ablation) | D-SCG localiser · 2D localisation | LSR: 0.265 -> 0.273 | Improved |
| use pretrained ConceptNet Numberbatch embeddings instead of learned (ablation) | D-SCG localiser · 2D localisation | LSR: 0.273 -> 0.297 | Improved |
| directed relative-position graph with end-to-end regression vs SCG-OL | D-SCG localiser · 2D localisation | LSR: 0.238 -> 0.297 | Improved |
| add AtLocation edges to proximity edges (ablation) | D-SCG localiser · edge type ablation | LSR: 0.257 -> 0.292 | Improved |
| add UsedFor edges to proximity edges (ablation) | D-SCG localiser · edge type ablation | LSR: 0.257 -> 0.272 | Improved |
| use both AtLocation and UsedFor edges (ablation) | D-SCG localiser · edge type ablation | LSR: 0.292 -> 0.297 | Improved |
| replace no-attention GraphSAGE with the proposed sparse attention (ablation) | D-SCG localiser · attention ablation | LSR: 0.199 -> 0.215 | Improved |
| replace GraphSAGE with GAT attention (ablation) | D-SCG localiser · attention ablation | LSR: 0.199 -> 0.179 | Hurt |
| replace GAT with GATv2 (ablation) | D-SCG localiser · attention ablation | LSR: 0.179 -> 0.202 | Improved |
| increase message passing layers from 1 to 4 (ablation) | D-SCG localiser · layers ablation | LSR: 0.180 -> 0.297 | Improved |
| increase message passing layers from 4 to 5 (ablation) | D-SCG localiser · layers ablation | LSR: 0.297 -> 0.285 | Hurt |
Closest papers by meaning
- Spatial Commonsense Graph for Object Localisation in Partial Scenes (CVPR 2022 · similarity 0.96)
- Learning Visual Commonsense for Robust Scene Graph Generation (ECCV 2020 · similarity 0.88)
- SG2Loc: Sequential Visual Localization on 3D Scene Graphs (ICML 2026 · similarity 0.87)
- ScanRefer: 3D Object Localization in RGB-D Scans using Natural Language (ECCV 2020 · similarity 0.87)
- From Objects to Anywhere: A Holistic Benchmark for Multi-level Visual Grounding in 3D Scenes (NeurIPS 2025 · similarity 0.85)
- Augmented Commonsense Knowledge for Remote Object Grounding (AAAI 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.
