Pin the Memory: Learning to Generalize Semantic Segmentation

TL;DR

This paper introduces a memory-guided meta-learning approach for semantic segmentation that enhances model generalization across unseen domains by abstracting class knowledge into a constant memory component.

Contribution

It proposes a novel memory-based meta-learning framework with specific loss functions to improve domain generalization in semantic segmentation tasks.

Findings

Outperforms state-of-the-art methods on multiple benchmarks.

Effectively learns domain-agnostic class representations.

Enhances segmentation accuracy in unseen domains.

Abstract

The rise of deep neural networks has led to several breakthroughs for semantic segmentation. In spite of this, a model trained on source domain often fails to work properly in new challenging domains, that is directly concerned with the generalization capability of the model. In this paper, we present a novel memory-guided domain generalization method for semantic segmentation based on meta-learning framework. Especially, our method abstracts the conceptual knowledge of semantic classes into categorical memory which is constant beyond the domains. Upon the meta-learning concept, we repeatedly train memory-guided networks and simulate virtual test to 1) learn how to memorize a domain-agnostic and distinct information of classes and 2) offer an externally settled memory as a class-guidance to reduce the ambiguity of representation in the test data of arbitrary unseen domain. To this end, we also propose memory divergence and feature cohesion losses, which encourage to learn memory reading and update processes for category-aware domain generalization. Extensive experiments for semantic segmentation demonstrate the superior generalization capability of our method over state-of-the-art works on various benchmarks.