Retrieval-Guided and Semantically Grounded Image Captioning for Open-Domain Scenes
Recent image captioning methods based on pre-trained vision–language models can generate fluent and coherent descriptions, yet they still struggle in open-domain scenes that contain long-tail concepts, uncommon object combinations, and ambiguous visual evidence. Two limitations are especially import...
Kaydedildi:
| Asıl Yazarlar: | , , , |
|---|---|
| Materyal Türü: | Artigo |
| Dil: | Inglês |
| Baskı/Yayın Bilgisi: |
MDPI AG
2026-05-01
|
| Seri Bilgileri: | Mathematics |
| Konular: | |
| Online Erişim: | https://www.mdpi.com/2227-7390/14/10/1667 |
| Etiketler: |
Etiket eklenmemiş, İlk siz ekleyin!
|