【Few-Shot Segmentation论文阅读笔记】Part-aware prototype for few-shot semantic Segmentation, ECCV, 2020
Abstract问题:现有Few-shotsegmentation方法的缺点包括:只能处理有限的问题:one-wayfew-shotsegmentation,比较难向multi-way进行扩展Singleprototype表征能力有限,无法涵盖object的全部regions目标:针对上述问题,本文引入semi-supervisedframework,将其作为semi-supervisedfew-