Enhancing Zero-shot Image Captioning through Image-like Retrieval and Frequency-based Entity Filtering
IFCap, a novel approach for zero-shot image captioning, addresses the modality gap between image and text data by performing Image-like Retrieval and integrating retrieved captions with input features through a Fusion Module. Additionally, it employs Frequency-based Entity Filtering to enhance caption quality by extracting frequently occurring entities from retrieved captions.
2024 NICE 이미지 캡셔닝 챌린지를 위한 솔루션
새로운 NICE 2024 데이터셋의 스타일과 내용의 차이를 해결하기 위해 검색 증강 및 캡션 등급 부여 방법을 통해 이미지 캡션을 효과적으로 향상시킴.
Enhancing Zero-Shot Image Captioning through Retrieval Augmentation and Caption-Level Strategies
A solution that leverages retrieval augmentation and caption-level strategies to effectively enhance zero-shot image captioning performance on the NICE 2024 dataset.
MeaCap: Memory-Augmented Zero-shot Image Captioning Framework
Proposing a novel Memory-Augmented zero-shot image Captioning framework (MeaCap) to generate concept-centered captions with high consistency and less hallucinations.
MeaCap: Memory-Augmented Zero-shot Image Captioning Framework
提案されたMeaCapフレームワークは、ゼロショット画像キャプショニングにおいて、記憶を活用して高品質なキャプション生成を実現する。
Memory-Augmented Zero-shot Image Captioning Framework: MeaCap
MeaCap proposes a novel Memory-Augmented framework for zero-shot image captioning, achieving state-of-the-art performance by integrating textual memory and visual-related fusion scores.
© 2024 by Linnk AI