Controllable and Diverse Image Captioning via Keyword-Driven Diffusion Networks
Image captioning is the task of understanding visual elements and their relationships in an image and expressing them in natural language.However, due to the ambiguity in image understanding and the diversity of language expressions, current image captioning models, which are primarily based on autoregressive methods, struggle to generate richer descriptions and often deviate from ICCIP 2024, November 21-24, 2024, Lingshui, China
Authors
- Simin Li (ORCID: https://orcid.org/0000-0001-6355-2978)
- Mingxuan Pan
- Shengjin Wang (ORCID: https://orcid.org/0000-0001-7809-1932)
- Yali Li (ORCID: https://orcid.org/0000-0002-6629-7228)
Institutions
- China Academy of Engineering Physics (CN)
- Tsinghua University (CN)
Publication Details
- Published
- 2024-11-21
- DOI
- https://doi.org/10.1145/3708657.3708748
- Primary Topic
- Multimodal Machine Learning Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00