Research map / Graph learning / Knowledge and temporal graphs
Sketching Image Gist: Human-Mimetic Hierarchical Scene Graph Generation
Wenbin Wang, Ruiping Wang, Shiguang Shan, Xilin Chen
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 |
|---|---|---|---|
| use IFS instead of AFS for choosing parent node (ablation) | HetH · VG150 PREDCLS | R@50: 64.7 -> 66.3 | Improved |
| add chain LSTM encoder to Hybrid-LSTM (ablation) | HetH · VG150 PREDCLS | R@50: 64.5 -> 66.3 | Improved |
| add chain LSTM encoder to Hybrid-LSTM (ablation) | HetH · VG150 SGCLS | R@50: 35.9 -> 36.6 | Improved |
| use IFS instead of AFS (ablation) | HetH · VG150 SGGEN | R@50: 27.1 -> 27.5 | Improved |
| use saliency and area maps jointly in RRM instead of neither (ablation) | HetH-RRM · VG-KR Triplet Match PREDCLS | kR@1: 16.2 -> 17.5 | Improved |
| add saliency map only to RRM (ablation) | RRM-Base · VG-KR Triplet Match PREDCLS | kR@1: 16.2 -> 17.2 | Improved |
| add area map only to RRM (ablation) | RRM-Base · VG-KR Triplet Match PREDCLS | kR@1: 16.2 -> 16.9 | Improved |
| add RRM to MOTIFS | MOTIFS · VG-KR Triplet Match PREDCLS | kR@1: 11.3 -> 16.7 | Improved |
| add RRM to HetH | HetH · VG-KR Triplet Match PREDCLS | kR@1: 11.6 -> 17.5 | Improved |
| rank relations with HetH-RRM instead of frequency for captioning input | HetH-Freq · VG-KR captioning (top relations 20) | CIDEr: 94.0 -> 102.2 | Improved |
| hierarchical multi-branch tree structure vs binary tree (VCTree-SL) | HetH · VG150 SGCLS | R@50: 37.9 -> 36.6 | Hurt |
Closest papers by meaning
- Unconditional Scene Graph Generation (ICCV 2021 · similarity 0.90)
- Scene Graph Generation by Iterative Message Passing (CVPR 2017 · similarity 0.90)
- A Comprehensive Survey of Scene Graphs: Generation and Application (TPAMI 2021 · similarity 0.90)
- Unbiased Heterogeneous Scene Graph Generation with Relation-Aware Message Passing Neural Network (AAAI 2023 · similarity 0.90)
- The Scene Language: Representing Scenes with Programs, Words, and Embeddings (CVPR 2025 · similarity 0.89)
- DSGG: Dense Relation Transformer for an End-to-end 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.
