Research map / Vision-language and multimodal / Human and embodied modeling
Knowledge-Enhanced Image Captioning with Adaptive Graph-based Multimodal Alignment and LLM
Guoyi Li, Die Hu, Haozhe Li, Zhongjiang Yao, Wei Mi, Zongzhen Liu, Xiaodan Zhang, Honglei Lyu
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 knowledge graph prompts (ablation) | AKGMA · UniKnowCap Test | RecogAcc: 48.0% -> 77.1% | Improved |
| add adaptive RL-based subgraph construction (AKGC) instead of random walk (ablation) | AKGMA · UniKnowCap Test | RecogAcc: 57.4% -> 77.1% | Improved |
| use AKGC instead of static 1-hop/2-hop neighbor expansion (ablation) | AKGMA · UniKnowCap Test | CHAIR-S: 8.3/8.9 -> 7.3 | Improved |
| add VKPairs training of EKA (ablation) | AKGMA · UniKnowCap Test | RecogAcc: 72.8% -> 77.1% | Improved |
| add SCKA and AKGC on top of EKA (ablation) | AKGMA · UniKnowCap Test | RecogAcc: 54.8% -> 77.1% | Improved |
| add scene-context knowledge adapter (SCKA) (ablation) | AKGMA · MSCOCO Test | CIDEr: 145.1 -> 149.3 | Improved |
| add knowledge graph prompts (ablation) | AKGMA · MSCOCO Test | CIDEr: 138.2 -> 149.3 | Improved |
| increase context vector length LD from 4 to 8 (ablation) | AKGMA · MSCOCO Test | CIDEr: 149.3 -> 148.5 | Hurt |
| decrease context vector length LD from 4 to 2 (ablation) | AKGMA · MSCOCO Test | CIDEr: 149.3 -> 144.7 | Hurt |
| add VKPairs training (ablation) | AKGMA · MSCOCO Test | CIDEr: 143.7 -> 149.3 | Improved |
| AKGMA vs best baseline on knowledge recognition (KnowCap) | AKGMA · KnowCap Test | CIDEr: 93.4 vs OFAK-Replay 99.6 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- EVCap — EVCap: Retrieval-Augmented Image Captioning with External Visual-Name Memory for Open-World Comprehension (CVPR 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MMKE-Bench: A Multimodal Editing Benchmark for Diverse Visual Knowledge (ICLR 2025 · similarity 0.89)
- Benchmarking Large Vision-Language Models via Directed Scene Graph for Comprehensive Image Captioning (CVPR 2025 · similarity 0.89)
- Enhancing Descriptive Captions with Visual Attributes for Multimodal Perception (CVPR 2026 · similarity 0.89)
- MMMG: A Massive, Multidisciplinary, Multi-Tier Generation Benchmark for Text-to-Image Reasoning (NeurIPS 2025 · similarity 0.88)
- LION: Empowering Multimodal Large Language Model with Dual-Level Visual Knowledge (CVPR 2024 · similarity 0.88)
- Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models (ICML 2026 · similarity 0.88)
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.
