Uncertainty Reduction for Model Adaptation in Semantic Segmentation

S, Prabhu Teja; Fleuret, François

doi:10.1109/cvpr46437.2021.00949

Cited by 30 publications

(10 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…zhang et al [49] present a new paradigm called DaC by combining the global class-wise pseudo labeling and the local neighborhood consistency. Besides, some works are tailored for the natural and medical image segmentation [50], [51], [52], [53], [54]. Despite the impressive progress these approaches have made for the image classification and segmentation tasks, they are not applicable for the detection tasks.…”

Section: Source Free Unsupervised Domain Adaptationmentioning

confidence: 99%

SGDA: Towards 3-D Universal Pulmonary Nodule Detection via Slice Grouped Domain Attention

Liu

Luo

et al. 2024

IEEE/ACM Trans. Comput. Biol. and Bioinf.

View full text Add to dashboard Cite

Lung cancer is the leading cause of cancer death worldwide. The best solution for lung cancer is to diagnose the pulmonary nodules in the early stage, which is usually accomplished with the aid of thoracic computed tomography (CT). As deep learning thrives, convolutional neural networks (CNNs) have been introduced into pulmonary nodule detection to help doctors in this labor-intensive task and demonstrated to be very effective. However, the current pulmonary nodule detection methods are usually domain-specific, and cannot satisfy the requirement of working in diverse real-world scenarios. To address this issue, we propose a slice grouped domain attention (SGDA) module to enhance the generalization capability of the pulmonary nodule detection networks. This attention module works in the axial, coronal, and sagittal directions. In each direction, we divide the input feature into groups, and for each group, we utilize a universal adapter bank to capture the feature subspaces of the domains spanned by all pulmonary nodule datasets. Then the bank outputs are combined from the perspective of domain to modulate the input group. Extensive experiments demonstrate that SGDA enables substantially better multi-domain pulmonary nodule detection performance compared with the state-of-the-art multi-domain learning methods.

show abstract

Section: Source Free Unsupervised Domain Adaptationmentioning

confidence: 99%

SGDA: Towards 3-D Universal Pulmonary Nodule Detection via Slice Grouped Domain Attention

Liu

Luo

et al. 2024

IEEE/ACM Trans. Comput. Biol. and Bioinf.

View full text Add to dashboard Cite

show abstract

“…According to the size of different datasets, we empirically set γ = 0.05, 0.1, 0.5, 0.1 for Office, Office-Home, VisDA, and DomainNet respectively. (Liang, Hu, and Feng 2020;Yang et al 2021;Liang et al 2022b), selfknowledge distillation (Chen et al 2022b), statistical alignment (Ishii and Sugiyama 2021), or uncertainty guidance (Sivaprasad and Fleuret 2021;Litrico, Del Bue, and Morerio 2023). Nonetheless, the issue of source data privacy and security still remains because the white-box source model is exposed to the target domain.…”

Section: Ablation Studiesmentioning

confidence: 99%

A Separation and Alignment Framework for Black-Box Domain Adaptation

Xia,

Zhao,

Lyu

et al. 2024

AAAI

View full text Add to dashboard Cite

Black-box domain adaptation (BDA) targets to learn a classifier on an unsupervised target domain while assuming only access to black-box predictors trained from unseen source data. Although a few BDA approaches have demonstrated promise by manipulating the transferred labels, they largely overlook the rich underlying structure in the target domain. To address this problem, we introduce a novel separation and alignment framework for BDA. Firstly, we locate those well-adapted samples via loss ranking and a flexible confidence-thresholding procedure. Then, we introduce a novel graph contrastive learning objective that aligns under-adapted samples to their local neighbors and well-adapted samples. Lastly, the adaptation is finally achieved by a nearest-centroid-augmented objective that exploits the clustering effect in the feature space. Extensive experiments demonstrate that our proposed method outperforms best baselines on benchmark datasets, e.g. improving the averaged per-class accuracy by 4.1% on the VisDA dataset. The source code is available at: https://github.com/MingxuanXia/SEAL.

show abstract

“…They rely on an entropy-based intra-domain module to leverage the correctly segmented patches as supervision during the model adaptation stage (see Figure 2.12). Sivaprasad and Fleuret (2021) propose a solution where the uncertainty of the target domain samples' predictions is minimized, while the robustness against noise perturbations in the feature space is maximized. Kundu et al (2021) decompose the problem into performing first source-only domain generalization and then adapting the model to the target by self-training with reliable target pseudo-labels.…”

Section: Source-free Domain Adaptationmentioning

confidence: 99%

Semantic Image Segmentation: Two Decades of Research

Volpi¹,

Chidlovskii²

2022

View full text Add to dashboard Cite

Semantic image segmentation (SiS) plays a fundamental role in a broad variety of computer vision applications, providing key information for the global understanding of an image. This survey is an effort to summarize two decades of research in the field of SiS, where we propose a literature review of solutions starting from early historical methods followed by an overview of more recent deep learning methods including the latest trend of using transformers. We complement the review by discussing particular cases of the weak supervision and side machine learning techniques that can be used to improve the semantic segmentation such as curriculum, incremental or self-supervised learning.State-of-the-art SiS models rely on a large amount of annotated samples, which are more expensive to obtain than labels for tasks such as image classification. Since unlabeled data is instead significantly cheaper to obtain, it is not surprising that Unsupervised Domain Adaptation (UDA) reached a broad success within the semantic segmentation community. Therefore, a second core contribution of this book is to summarize five years of a rapidly growing field, Domain Adaptation for Semantic Image Segmentation (DASiS) which embraces the importance of semantic segmentation itself and a critical need of adapting segmentation models to new environments. In addition to providing a comprehensive survey on DASiS techniques, we unveil also newer trends

show abstract

Uncertainty Reduction for Model Adaptation in Semantic Segmentation

Cited by 30 publications

References 50 publications

SGDA: Towards 3-D Universal Pulmonary Nodule Detection via Slice Grouped Domain Attention

SGDA: Towards 3-D Universal Pulmonary Nodule Detection via Slice Grouped Domain Attention

A Separation and Alignment Framework for Black-Box Domain Adaptation

Semantic Image Segmentation: Two Decades of Research

Contact Info

Product

Resources

About