zero-shot classification
Zero-shot classification is a machine learning approach that enables models to recognize classes of which they have not seen any training instances, often utilizing learned semantic relationships between classes.
- AmorLIP: Efficient Language-Image Pretraining via Amortization
- Enhancing CLIP Robustness via Cross-Modality Alignment
- ExGra-Med: Extended Context Graph Alignment for Medical Vision-Language Models
- Explaining Similarity in Vision-Language Encoders with Weighted Banzhaf Interactions
- Learning Shared Representations from Unpaired Data
- Meta CLIP 2: A Worldwide Scaling Recipe
- Noise Matters: Optimizing Matching Noise for Diffusion Classifiers
- Perception Encoder: The best visual embeddings are not at the output of the network
- Perception Encoder: The best visual embeddings are not at the output of the network
- SceneForge: Enhancing 3D-text alignment with Structured Scene Compositions
- Training-Free Test-Time Adaptation via Shape and Style Guidance for Vision-Language Models
- Unlabeled Data Improves Fine-Grained Image Zero-shot Classification with Multimodal LLMs
- With Limited Data for Multimodal Alignment, Let the STRUCTURE Guide You