Few-shot segmentation (FSS) methods perform image segmentation for a\nparticular object class in a target (query) image, using a small set of\n(support) image-mask pairs. Recent deep neural network based FSS methods\nleverage high-dimensional feature similarity between the foreground features of\nthe support images and the query image features. In this work, we demonstrate\ngaps in the utilization of this similarity information in existing methods, and\npresent a framework - SimPropNet, to bridge those gaps. We propose to jointly\npredict the support and query masks to force the support features to share\ncharacteristics with the query features. We also propose to utilize\nsimilarities in the background regions of the query and support images using a\nnovel foreground-background attentive fusion mechanism. Our method achieves\nstate-of-the-art results for one-shot and five-shot segmentation on the\nPASCAL-5i dataset. The paper includes detailed analysis and ablation studies\nfor the proposed improvements and quantitative comparisons with contemporary\nmethods.\n