Conditioned and composed image retrieval combining and partially fine-tuning CLIP-based features.
Alberto BaldratiMarco BertiniTiberio UricchioAlberto Del BimboPublished in: CVPR Workshops (2022)
Keyphrases
- fine tuning
- image retrieval
- feature space
- low level
- feature set
- feature vectors
- image features
- image database
- video clips
- classification accuracy
- low level features
- fine tune
- viable alternative
- extracted features
- machine learning
- key frames
- image representation
- support vector
- feature selection
- image retrieval systems
- data sets