Exploring Test-Time Adaptation for Object Detection in Continually Changing Environments

TL;DR

This paper introduces AMROD, a novel test-time adaptation method for object detection in changing environments, improving pseudo-label quality and knowledge retention through object-level contrastive learning, adaptive thresholds, and selective parameter restoration.

Contribution

AMROD presents a comprehensive approach combining contrastive learning, adaptive thresholding, and selective parameter resetting to enhance continual test-time adaptation for object detection.

Findings

AMROD outperforms existing methods in CTTA object detection tasks.

Achieves 3.2 mAP improvement on Cityscapes-to-Cityscapes-C.

Increases efficiency by 20% in the CTTA setting.

Abstract

Real-world application models are commonly deployed in dynamic environments, where the target domain distribution undergoes temporal changes. Continual Test-Time Adaptation (CTTA) has recently emerged as a promising technique to gradually adapt a source-trained model to continually changing target domains. Despite recent advancements in addressing CTTA, two critical issues remain: 1) Fixed thresholds for pseudo-labeling in existing methodologies lead to low-quality pseudo-labels, as model confidence varies across categories and domains; 2) Stochastic parameter restoration methods for mitigating catastrophic forgetting fail to preserve critical information effectively, due to their intrinsic randomness. To tackle these challenges for detection models in CTTA scenarios, we present AMROD, featuring three core components. Firstly, the object-level contrastive learning module extracts object-level features for contrastive learning to refine the feature representation in the target domain. Secondly, the adaptive monitoring module dynamically skips unnecessary adaptation and updates the category-specific threshold based on predicted confidence scores to enable efficiency and improve the quality of pseudo-labels. Lastly, the adaptive randomized restoration mechanism selectively reset inactive parameters with higher possibilities, ensuring the retention of essential knowledge. We demonstrate the effectiveness of AMROD on four CTTA object detection tasks, where AMROD outperforms existing methods, especially achieving a 3.2 mAP improvement and a 20\% increase in efficiency on the Cityscapes-to-Cityscapes-C CTTA task. The code of this work is available at https://github.com/ShileiCao/AMROD.