Universal Domain Adaptive Object Detector

TL;DR

This paper introduces US-DAF, a novel universal domain adaptive object detection method that effectively handles category and scale shifts across domains, achieving state-of-the-art results on multiple benchmarks.

Contribution

The paper proposes a new framework with filter and scale-aware modules to address category and scale shifts in universal domain adaptive object detection.

Findings

Achieves state-of-the-art results on three scenarios

Yields 7.1% and 5.9% improvements on Clipart1k and Watercolor datasets

Effectively reduces negative transfer and improves transferability

Abstract

Universal domain adaptive object detection (UniDAOD)is more challenging than domain adaptive object detection (DAOD) since the label space of the source domain may not be the same as that of the target and the scale of objects in the universal scenarios can vary dramatically (i.e, category shift and scale shift). To this end, we propose US-DAF, namely Universal Scale-Aware Domain Adaptive Faster RCNN with Multi-Label Learning, to reduce the negative transfer effect during training while maximizing transferability as well as discriminability in both domains under a variety of scales. Specifically, our method is implemented by two modules: 1) We facilitate the feature alignment of common classes and suppress the interference of private classes by designing a Filter Mechanism module to overcome the negative transfer caused by category shift. 2) We fill the blank of scale-aware adaptation in object detection by introducing a new Multi-Label Scale-Aware Adapter to perform individual alignment between the corresponding scale for two domains. Experiments show that US-DAF achieves state-of-the-art results on three scenarios (i.e, Open-Set, Partial-Set, and Closed-Set) and yields 7.1% and 5.9% relative improvement on benchmark datasets Clipart1k and Watercolor in particular.