Multi-Head Distillation for Continual Unsupervised Domain Adaptation in Semantic Segmentation

TL;DR

This paper introduces MuHDi, a multi-head distillation framework for continual unsupervised domain adaptation in semantic segmentation, effectively handling multiple sequentially discovered target domains without forgetting previous ones.

Contribution

The work presents a novel multi-head distillation approach that addresses catastrophic forgetting in continual UDA for semantic segmentation, enabling models to adapt to multiple target domains sequentially.

Findings

MuHDi outperforms baseline methods on multi-target UDA benchmarks.

The approach effectively mitigates catastrophic forgetting in continual learning.

Extensive ablation studies validate the architecture's components.

Abstract

Unsupervised Domain Adaptation (UDA) is a transfer learning task which aims at training on an unlabeled target domain by leveraging a labeled source domain. Beyond the traditional scope of UDA with a single source domain and a single target domain, real-world perception systems face a variety of scenarios to handle, from varying lighting conditions to many cities around the world. In this context, UDAs with several domains increase the challenges with the addition of distribution shifts within the different target domains. This work focuses on a novel framework for learning UDA, continuous UDA, in which models operate on multiple target domains discovered sequentially, without access to previous target domains. We propose MuHDi, for Multi-Head Distillation, a method that solves the catastrophic forgetting problem, inherent in continual learning tasks. MuHDi performs distillation at multiple levels from the previous model as well as an auxiliary target-specialist segmentation head. We report both extensive ablation and experiments on challenging multi-target UDA semantic segmentation benchmarks to validate the proposed learning scheme and architecture.