AIR-DA: Adversarial Image Reconstruction for Unsupervised Domain Adaptive Object Detection

TL;DR

This paper introduces AIR-DA, a novel adversarial image reconstruction method that improves unsupervised domain adaptive object detection by balancing adversarial training and enhancing feature alignment, leading to superior performance across various datasets.

Contribution

It proposes a new regularization technique, AIR, and a multi-level feature alignment module to improve domain adaptation in object detection tasks.

Findings

Outperforms previous methods on multiple domain shift datasets.

Balances adversarial training between feature extractor and discriminator.

Enhances domain-invariance in feature representations.

Abstract

Unsupervised domain adaptive object detection is a challenging vision task where object detectors are adapted from a label-rich source domain to an unlabeled target domain. Recent advances prove the efficacy of the adversarial based domain alignment where the adversarial training between the feature extractor and domain discriminator results in domain-invariance in the feature space. However, due to the domain shift, domain discrimination, especially on low-level features, is an easy task. This results in an imbalance of the adversarial training between the domain discriminator and the feature extractor. In this work, we achieve a better domain alignment by introducing an auxiliary regularization task to improve the training balance. Specifically, we propose Adversarial Image Reconstruction (AIR) as the regularizer to facilitate the adversarial training of the feature extractor. We further design a multi-level feature alignment module to enhance the adaptation performance. Our evaluations across several datasets of challenging domain shifts demonstrate that the proposed method outperforms all previous methods, of both one- and two-stage, in most settings.