Research map / Graph learning / Knowledge and temporal graphs
LinkNet: Relational Embedding for Scene Graph
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 relational embedding over object instances | Faster R-CNN scene graph model · Visual Genome, Scene Graph Classification | R@100 | Helped |
| add global context encoding and geometrical layout encoding together to REM (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 40.8 -> 41.7 | Helped |
| add GLEM to REM (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 40.8 -> 41.2 | Helped |
| add GCEM to REM (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 40.8 -> 41.3 | Helped |
| use 2 REMs instead of 1 (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 41 -> 41.7 | Helped |
| increase REM count from 2 to 3 or 4 (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 41.7 -> 41.3; 41.7 -> 41.4 | Hurt |
| set reduction ratio to 1 rather than 2 (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 41.7 -> 41.6 | Hurt |
| set reduction ratio to 4 rather than 2 (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 41.7 -> 41.6 | Hurt |
| set reduction ratio to 8 rather than 2 (ablation) | LinkNet · Visual Genome, Scene Graph Classification | R@100: 41.7 -> 41.2 | Hurt |
| include both argmax(O4) and concat(O3) in E0 (ablation) | LinkNet edge-relational embedding · Visual Genome, Scene Graph Classification | R@100: 40.6 -> 41.7; 41.4 -> 41.7 | Helped |
| use row-wise softmax rather than sigmoid (ablation) | LinkNet relational embedding · Visual Genome, Scene Graph Classification | R@100: 41 -> 41.7 | Helped |
| use dot product rather than Euclidean distance (ablation) | LinkNet relational embedding · Visual Genome, Scene Graph Classification | R@100: 40.7 -> 40.8 | Helped |
Closest papers by meaning
- Scene Graph Generation by Iterative Message Passing (CVPR 2017 · similarity 0.93)
- Attentive Relational Networks for Mapping Images to Scene Graphs (CVPR 2019 · similarity 0.92)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.91)
- Scene Graph-Grounded Image Generation (AAAI 2025 · similarity 0.91)
- Object-Centric Representation Learning for Enhanced 3D Semantic Scene Graph Prediction (NeurIPS 2025 · similarity 0.90)
- EGTR: Extracting Graph from Transformer for Scene Graph Generation (CVPR 2024 · similarity 0.89)
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.