Structured Query-Based Image Retrieval Using Scene Graphs

CVPR 2020

Authors: Brigit Schroeder, Subarna Tripathi Description: A structured query can capture the complexity of object interactions in images (e.g. 'woman rides motorcycle') unlike single objects (e.g. 'woman' or 'motorcycle'). Image retrieval using structured queries therefore is much more useful than single object retrieval, but a much more challenging problem. In this paper we present a method which uses scene graph embeddings as the basis of our image retrieval approach. We examine how visual relationships, derived from scene graphs, can be used as structured queries. Notably, we are able to achieve high recall even on low to medium frequency objects found in the long-tailed COCO-Stuff dataset, and find adding a visual relationship-inspired loss boosts our recall by 10% in the best case.