TL;DR使用卷积神经网络(CNN)对支持图像进行特征提取,生成类别特征向量,然后通过余弦相似度将目标物体在查询图像中分割出来,在 PASCAL-$5^i$ and COCO-$20^i$ 数据集上验证了其优越性。
Abstract
This paper is about few-shot segmentation of foreground objects in images. We
train a cnn on small subsets of training images, each mimicking the few-shot
setting. In each subset, one image serves as the query an