M2-RAAP: A Multi-Modal Recipe for Advancing Adaptation-based Pre-training towards Effective and Efficient Zero-shot Video-text Retrieval.
Xingning DongZipeng FengChunluan ZhouXuzheng YuMing YangQingpei GuoPublished in: CoRR (2024)
Keyphrases
- multi modal
- text retrieval
- semantic concepts
- video search
- multi modality
- information retrieval
- inverted file
- document retrieval
- video sequences
- retrieval systems
- image retrieval
- document collections
- audio visual
- multimedia
- multimedia retrieval
- retrieval model
- query expansion
- cross modal
- video database
- multimedia information retrieval
- high dimensional
- multiple modalities
- key frames
- feature vectors
- training set