Xinya Du

Papers

Total Citations

H-Index

About

Xinya Du is a rising researcher at the forefront of vision-language learning and few-shot classification. Their most impactful contribution is the development of **Proto-CLIP**, a novel framework that bridges large-scale vision-language models like CLIP with prototypical networks. By introducing dual image and text prototypes, Du’s work enables models to learn from as few as one or two examples—a critical advance for data-scarce domains. The Proto-CLIP papers, published in 2023 and 2024, have already garnered **18 total citations**, signaling strong early interest from the community. This work directly addresses a fundamental challenge in AI: how to generalize from limited supervision. Du’s approach is notable for its elegant fusion of multimodal pretraining and metric learning, offering a practical path to few-shot recognition without requiring extensive fine-tuning. As the field pushes toward more efficient and generalizable models, Xinya Du’s research stands out for its clarity of design and immediate applicability. With Proto-CLIP laying a strong foundation, Du is well-positioned to make further strides in low-shot learning and multimodal understanding.