PixMatch: Unsupervised Domain Adaptation via Pixelwise Consistency Training

TL;DR

PixMatch introduces a pixelwise consistency training framework for unsupervised domain adaptation in semantic segmentation, leveraging input perturbations to improve model performance on real-world data without complex adversarial methods.

Contribution

The paper proposes a simple, effective, and memory-efficient pixelwise consistency loss for unsupervised domain adaptation, outperforming adversarial approaches on synthetic-to-real benchmarks.

Findings

Achieves strong results on GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes benchmarks.

Simpler and more memory-efficient than adversarial adaptation methods.

Demonstrates the effectiveness of input perturbation-based consistency training.

Abstract

Unsupervised domain adaptation is a promising technique for semantic segmentation and other computer vision tasks for which large-scale data annotation is costly and time-consuming. In semantic segmentation, it is attractive to train models on annotated images from a simulated (source) domain and deploy them on real (target) domains. In this work, we present a novel framework for unsupervised domain adaptation based on the notion of target-domain consistency training. Intuitively, our work is based on the idea that in order to perform well on the target domain, a model's output should be consistent with respect to small perturbations of inputs in the target domain. Specifically, we introduce a new loss term to enforce pixelwise consistency between the model's predictions on a target image and a perturbed version of the same image. In comparison to popular adversarial adaptation methods, our approach is simpler, easier to implement, and more memory-efficient during training. Experiments and extensive ablation studies demonstrate that our simple approach achieves remarkably strong results on two challenging synthetic-to-real benchmarks, GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes. Code is available at: https://github.com/lukemelas/pixmatch