Research map / Graph learning / Knowledge and temporal graphs
Auto-Encoding Scene Graphs for Image Captioning
Check 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 the SGAE scene-graph auto-encoder and shared dictionary to the baseline captioner (ablation) | Base · MS-COCO Karpathy split | CIDEr-D: 120.6 -> 127.8 | Helped |
| Add MGCN (ablation) | Base · MS-COCO Karpathy split | CIDEr-D: 120.6 -> 123.4 | Helped |
| Add dictionary learning without GCN or MGCN (ablation) | Base · MS-COCO Karpathy split | CIDEr-D: 120.6 -> 124.2 | Helped |
| Add GCN-trained shared dictionary (ablation) | Base · MS-COCO Karpathy split | CIDEr-D: 120.6 -> 125.7 | Helped |
| Add MGCN to the dictionary-based model (ablation) | Base+D · MS-COCO Karpathy split | CIDEr-D: 125.7 -> 127.8 | Helped |
| Learn the dictionary from Visual Genome text rather than matched MS-COCO captions (ablation) | Base captioner with shared dictionary · MS-COCO Karpathy split | CIDEr-D: 120.6 -> 123.2; 123.2 -> 127.8 | Mixed |
| Apply dictionary re-encoding during sentence reconstruction (ablation) | SGAE sentence reconstruction · Sentence reconstruction | CIDEr-D: 177.0 -> 153.2 | Hurt |
| Use SGAE instead of GCN-LSTM | Single-model captioning · MS-COCO Karpathy split | CIDEr-D: 127.6 -> 127.8 | Helped |
| Fuse model predictions | SGAE · MS-COCO Karpathy split | CIDEr-D: 127.8 -> 129.1 | Helped |
| Use SGAE instead of CAVP | Single-model captioning · MS-COCO official test server (c40) | CIDEr-D: 123.8 -> 125.5 | Helped |
Closest papers by meaning
- Benchmarking Large Vision-Language Models via Directed Scene Graph for Comprehensive Image Captioning (CVPR 2025 · similarity 0.87)
- SGDiff: Scene Graph Guided Diffusion Model for Image Collaborative SegCaptioning (AAAI 2025 · similarity 0.87)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · similarity 0.86)
- LLM4SGG: Large Language Models for Weakly Supervised Scene Graph Generation (CVPR 2024 · similarity 0.86)
- Panoptic Scene Graph Generation with Semantics-Prototype Learning (AAAI 2024 · similarity 0.86)
- SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025 · 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.