The author presents the Query-guided Prototype Evolution Network (QPENet) as a novel method to enhance Few-Shot Segmentation by integrating query features into prototype generation, resulting in customized solutions tailored to specific queries.
Query-guided Prototype Evolution Network (QPENet) integrates query features to create customized prototypes for improved few-shot segmentation.
Query-guided Prototype Evolution Network (QPENet) integrates support and query features to create customized prototypes for precise segmentation.
Introducing the Query-guided Prototype Evolution Network (QPENet) to enhance Few-Shot Segmentation by integrating query features into prototype generation.
A novel prompt-driven scheme called "Prompt and Transfer" (PAT) is proposed to dynamically tune the encoder for focusing on class-specific objects in different Few-shot Segmentation tasks.
본 논문은 프롬프트 기반 동적 클래스 인식 기법을 제안하여 적은 수의 지원 이미지로도 효과적으로 미지의 클래스를 세분화할 수 있다.
지원 정보의 효과적인 활용을 위해 지원 정보 재활용 마바와 쿼리 정보 차단 마바를 결합한 하이브리드 마바 네트워크를 제안한다.
This paper introduces a novel inference-time pseudo-labeling technique to improve the performance of few-shot 3D medical image segmentation models, particularly in addressing the challenge of limited annotated data.