On Class Imbalance and Background Filtering in Visual Relationship Detection
This work addresses critical limitations in VRD models for computer vision applications, but it is incremental as it builds on existing methods to improve specific bottlenecks.
The paper tackles class imbalance and background filtering issues in Visual Relationship Detection (VRD), where models struggle with uncommon classes and irrelevant relationships, proposing modifications to models and training to address these problems and suggesting new evaluation measures.
In this paper we investigate the problems of class imbalance and irrelevant relationships in Visual Relationship Detection (VRD). State-of-the-art deep VRD models still struggle to predict uncommon classes, limiting their applicability. Moreover, many methods are incapable of properly filtering out background relationships while predicting relevant ones. Although these problems are very apparent, they have both been overlooked so far. We analyse why this is the case and propose modifications to both model and training to alleviate the aforementioned issues, as well as suggesting new measures to complement existing ones and give a more holistic picture of the efficacy of a model.