MedCLIP:

Zifeng Wang1, Zhenbang Wu1, Dinesh Agarwal1,2

  • 1Department of Computer Science, University of Illinois Urbana-Champaign.

Proceedings of the Conference on Empirical Methods in Natural Language Processing. Conference on Empirical Methods in Natural Language Processing
|August 15, 2024
PubMed
概括

通过将图像和文本脱而出,MedCLIP增强了医疗视觉-文本学习,大大减少了虚假负面. 这种方法以更少的数据实现了最先进的结果,改善了零射击预测和检索.