Fine-tuned Vision Transformers for near-duplicate image retrieval. Fine-tuning methodology: contrastive learning with multisimilarity loss and positive/negative augmentations.
Paper with training details and metrics to be uploaded soon.
- Downloads last month
- 57
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for rfrade/near_duplicate_retrieval_vit
Base model
google/vit-base-patch16-224