What is Right for Me is Not Yet Right for You: A Dataset for Grounding Relative Directions via Multi-Task Learning

TL;DR

This paper introduces GRiD-3D, a new dataset for understanding relative spatial directions in images, and demonstrates that neural networks can learn to ground these directions through multi-task learning.

Contribution

The paper presents a novel dataset for relative directions and analyzes how end-to-end models can learn to ground these spatial relations.

Findings

Models can learn to answer relative direction questions.

Subtasks are learned in an order reflecting an intuitive processing pipeline.

Grounding relative directions is feasible with appropriate training.

Abstract

Understanding spatial relations is essential for intelligent agents to act and communicate in the physical world. Relative directions are spatial relations that describe the relative positions of target objects with regard to the intrinsic orientation of reference objects. Grounding relative directions is more difficult than grounding absolute directions because it not only requires a model to detect objects in the image and to identify spatial relation based on this information, but it also needs to recognize the orientation of objects and integrate this information into the reasoning process. We investigate the challenging problem of grounding relative directions with end-to-end neural networks. To this end, we provide GRiD-3D, a novel dataset that features relative directions and complements existing visual question answering (VQA) datasets, such as CLEVR, that involve only absolute directions. We also provide baselines for the dataset with two established end-to-end VQA models. Experimental evaluations show that answering questions on relative directions is feasible when questions in the dataset simulate the necessary subtasks for grounding relative directions. We discover that those subtasks are learned in an order that reflects the steps of an intuitive pipeline for processing relative directions.