Position: Do Not Explain Vision Models Without Context

TL;DR

This paper highlights the importance of incorporating spatial context into explainable AI methods for vision models, demonstrating current limitations and proposing new research directions for more effective explanations.

Contribution

It reviews existing explanation methods, identifies their neglect of context, and suggests new approaches emphasizing 'how' explanations should incorporate spatial relationships.

Findings

Current explanation methods often ignore spatial context.

Contextual information can significantly improve explanation quality.

Proposes new research directions for context-aware explanations.

Abstract

Does the stethoscope in the picture make the adjacent person a doctor or a patient? This, of course, depends on the contextual relationship of the two objects. If it's obvious, why don't explanation methods for vision models use contextual information? In this paper, we (1) review the most popular methods of explaining computer vision models by pointing out that they do not take into account context information, (2) show examples of failures of popular XAI methods, (3) provide examples of real-world use cases where spatial context plays a significant role, (4) propose new research directions that may lead to better use of context information in explaining computer vision models, (5) argue that a change in approach to explanations is needed from 'where' to 'how'.