Research map / Graph learning / Knowledge and temporal graphs
Graph reasoning for visual recognition
Model relations among regions, actors, and skeleton joints using graph networks.
96 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Knowledge and temporal graphs, keyword cluster “scene · image · visual”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
14 papers in 2023–24 and 13 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 5.4% to 4.6%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2017 · The More You Know: Using Knowledge Graphs for Image Classification (CVPR 2017 · 338 citations)
- 2018 · Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition (AAAI 2018 · 5,095 citations)
- 2019 · Two-Stream Adaptive Graph Convolutional Networks for Skeleton-Based Action Recognition (CVPR 2019 · 2,091 citations)
- 2020 · Skeleton-Based Action Recognition With Shift Graph Convolutional Network (CVPR 2020 · 996 citations)
- 2021 · Learning Graph Embeddings for Compositional Zero-Shot Learning (CVPR 2021 · 167 citations)
- 2022 · Vision GNN: An Image is Worth Graph of Nodes (NeurIPS 2022 · 192 citations)
- 2023 · 3Mformer: Multi-Order Multi-Mode Transformer for Skeletal Action Recognition (CVPR 2023 · 92 citations)
- 2024 · BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · 176 citations)
- 2025 · Towards Satellite Image Road Graph Extraction: A Global-Scale Dataset and A Novel Method (CVPR 2025 · 17 citations)
- 2026 · Graph Neural Field with Spatial-Correlation Augmentation for HRTF Personalization (AAAI 2026 · 2 citations)
Earliest papers here
- The More You Know: Using Knowledge Graphs for Image Classification (CVPR 2017 · 338 citations)
- Pixels to Graphs by Associative Embedding (NeurIPS 2017 · 205 citations)
- Learning Patch-Based Dynamic Graph for Visual Tracking (AAAI 2017 · 36 citations)
What has been tried
90 papers here report 1,013 measured changes (a component added, removed or swapped, compared with a baseline): helped 78%, no clear effect 4%, hurt 15%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Replace vanilla GC with BlockGC | NTU RGB+D 120 X-Sub · accuracy | 85.2 -> 85.8 | Helped | BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · 176 citations) |
| Add node- and edge-type semantic modules | NTU60-XSub · accuracy | 90.0 -> 90.8 | Helped | Dynamic Semantic-Based Spatial Graph Convolution Network for Skeleton-Based Human Action Recognition (AAAI 2024 · 58 citations) |
| add Local Sign Graph (LSG) module on ResNet18 baseline | PHOENIX14T dev · WER | decreases by 3.1% (LSG1) | Helped | SignGraph: A Sign Sequence is Worth Graphs of Nodes (CVPR 2024 · 28 citations) |
| Add node-guided resampling | SpaceNet · F1 | 80.89 -> 81.57 | Helped | Towards Satellite Image Road Graph Extraction: A Global-Scale Dataset and A Novel Method (CVPR 2025 · 17 citations) |
| Use partially connected composite-node graph instead of hierarchical neighboring-frame fully connected graph | MOT17 validation, GT detections · IDF1 | 76.2 -> 76.3 | Helped | Multi-Scene Generalized Trajectory Global Graph Solver with Composite Nodes for Multiple Object Tracking (AAAI 2024 · 17 citations) |
| add visual features to textual features | Baby · R@20 | 0.1010 -> 0.1016 | Helped | Beyond Graph Convolution: Multimodal Recommendation with Topology-aware MLPs (AAAI 2025 · 10 citations) |
| Add Dynamic Hypergraph Convolution, replacing Feature Dispatching | ImageNet-1K · Top-1 accuracy | +1.1 | Helped | DVHGNN: Multi-Scale Dilated Vision HGNN for Efficient Vision Recognition (CVPR 2025 · 8 citations) |
| use randomized rather than ordered GT key point prompts | OpenLane-V2 subset_A lane segment ablation · TOP_lsls | 25.1 -> 27.8 | Helped | Topo2Seq: Enhanced Topology Reasoning via Topology Sequence Learning (AAAI 2025 · 3 citations) |
| add GT to direct prototype learning | BIWI-S · mAP | 11.3 -> 26.7 | Helped | Motif Guided Graph Transformers with Combinatorial Skeleton Prototype Learning for Skeleton-Based Person Re-Identification (AAAI 2025 · 2 citations) |
| add image-graph block to Swin-B | EyePACS · Kappa | 88.2 -> 89.0 | Helped | SIGraph: Saliency Image-Graph Network for Retinal Disease Classification in Fundus Image (AAAI 2025 · 1 citation) |
| Remove the MLP decoder | PSG test · NDCG@3 | 28.27 -> 24.95 | Hurt | SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025) |
| apply rotation to each tile | Full-LGP, high thresholds · APLS | 0.187 -> 0.208 | Helped | Bezier Everywhere All at Once: Learning Drivable Lanes as Bezier Graphs (CVPR 2024) |
What reviewers will expect
Datasets and baselines named in the 8 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- ImageNet-1K · 2 of 8 papers
Baselines
- InceptionNeXt · 2 of 8 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 27 papers; from OpenAlex and the papers' affiliations).
- Zhejiang University · 2 papers
- Tencent · 2 papers
- Chinese Academy of Sciences · 2 papers
- Stanford University · 2 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Graph-based dialog systems (94 papers)
- Heterogeneous graph learning (66 papers)
- Social and behavior graphs (87 papers)
- Scene graph generation (70 papers)
Elsewhere in the topic
- Continuous-time dynamic graph learning (132 papers)
- Knowledge graph embeddings (110 papers)
- Temporal knowledge graph reasoning (84 papers)
- Temporal graph neural architectures (69 papers)
- Logical query answering and rule reasoning (68 papers)
- Spatio-temporal graph traffic forecasting (60 papers)
- Inductive knowledge graph completion (38 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).