Comprehensive Multi-Modal Interactions for Referring Image Segmentation

We investigate Referring Image Segmentation (RIS), which outputs a segmentation map corresponding to the given natural language description. To solve RIS efficiently, we need to understand each word's relationship with other words, each region in the image to other regions, and cross-modal alignment between linguistic and visual domains. We argue that one of the limiting factors in the recent methods is that they do not handle these interactions simultaneously. To this end, we propose a novel architecture called JRNet, which uses a Joint Reasoning Module(JRM) to concurrently capture the inter-modal and intra-modal interactions. The output of JRM is passed through a novel Cross-Modal Multi-Level Fusion (CMMLF) module which further refines the segmentation masks by exchanging contextual information across visual hierarchy through linguistic features acting as a bridge. We present thorough ablation studies and validate our approach's performance on four benchmark datasets, showing considerable performance gains over the existing state-of-the-art methods.