Unbiased Mean Teacher for Cross-domain Object Detection

By Jinhong Deng, Wen Li, Yuhua Chen, Lixin Duan. 2020

400 citations recorded by the source corpus.

Cross-domain object detection is challenging, because object detection model is often vulnerable to data variance, especially to the considerable domain shift between two distinctive domains. In this paper, we propose a new Unbiased Mean Teacher (UMT) model for cross-domain object detection. We reveal that there often exists a considerable model bias for the simple mean teacher (MT) model in cross-domain scenarios, and eliminate the model bias with several simple yet highly effective strategies. In particular, for the teacher model, we propose a cross-domain distillation method for MT to maximally exploit the expertise of the teacher model. Moreover, for the student model, we alleviate its bias by augmenting training samples with pixel-level adaptation. Finally, for the teaching process, we employ an out-of-distribution estimation strategy to select samples that most fit the current model to further enhance the cross-domain distillation process. By tackling the model bias issue with these strategies, our UMT model achieves mAPs of 44.1%, 58.1%, 41.7%, and 43.1% on benchmark datasets Clipart1k, Watercolor2k, Foggy Cityscapes, and Cityscapes, respectively, which outperforms the existing state-of-the-art results in notable margins. Our implementation is available at https://github.com/kinredon/umt.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Federated Class-Incremental Learning (2022, background): “Different from existing knowledge distillation strategies [1, 3, 8, 17, 26] that only ensure old classes’ semantic consistency among Θt−1 l and Θ r,t l , we consider inter-class relations between old and new classes simultaneously via optimizing LRD.”
Cross-Domain Adaptive Teacher for Object Detection (2021, methodology): “We compare our method with SCL [35], SWDA [32], DM [19], CRDA [42], HTCN [2], UMT [8], Source (F-RCNN), and Oracle (F-RCNN).”
BAPA-Net: Boundary Adaptation and Prototype Alignment for Cross-domain Semantic Segmentation (2021, methodology): “A variety of unsupervised domain adaptation methods are proposed to maximally eliminate the domain discrepancy through adversarial feature learning [22, 6, 13], entropy minimization [52, 53, 5], selftraining [65, 66, 31, 11], etc.”
MIC: Masked Image Consistency for Context-Enhanced Domain Adaptation (2022, methodology): “8 following [15,53] and the quality confidence threshold τ=0.”
Denoised Maximum Classifier Discrepancy for Source-Free Unsupervised Domain Adaptation (2022, methodology): “While many methods have been proposed to solve the UDA problem (Ganin and Lempitsky 2015; Long et al. 2018; Saito et al. 2018; Dong et al. 2020; Liu et al. 2021; Deng et al. 2021; Dong et al. 2021), these approaches is needed to access the source domain data during the training process.”
Undoing the Damage of Label Shift for Cross-domain Semantic Segmentation (2022, background): “, classification [28, 39], object detection [10, 15], and semantic segmentation [23, 42].”