Related references
Note: Only part of the references are listed.
Article
Computer Science, Artificial Intelligence
Matthias De Lange et al.
Summary: This article introduces the application of artificial neural networks in continual learning, focusing on task incremental classification. It proposes a new framework for continually evaluating the stability-plasticity trade-off of the network and performs experimental comparisons of 11 state-of-the-art continual learning methods, evaluating their strengths and weaknesses by considering different benchmark datasets.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Neerav Karani et al.
Summary: In this study, a method to address the mismatch between training and test images in medical image segmentation is proposed. The segmentation network consists of a shallow image normalization CNN and a deep CNN, with a denoising autoencoder used to model an implicit prior. Experimental results demonstrate the performance improvement achieved by test-time adaptation.
MEDICAL IMAGE ANALYSIS
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Jogendra Nath Kundu et al.
Summary: This study introduces a method for source-free domain adaptation in semantic segmentation by partitioning the task into source-only domain generalization and source-free target adaptation. The approach involves training a multi-head framework and utilizing reliable target pseudo-labels for self-training, as well as using a novel auto-encoder to enhance pseudo-label quality. Experimental results demonstrate the superiority of the proposed method compared to prior-arts and its compatibility with online adaptation for deployment in changing environments.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Viraj Prabhu et al.
Summary: SENTRY is a UDA algorithm that selectively minimizes or maximizes predictive entropy based on the predictive consistency of target instances under a committee of random image transformations, leading to significant improvements over existing methods in various benchmarks when combined with pseudolabel-based approximate target class balancing.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Christos Sakaridis et al.
Summary: The ACDC dataset consists of 4006 images equally distributed among four common adverse conditions, namely fog, nighttime, rain, and snow, for training and testing semantic segmentation methods. Each image is accompanied by high-quality fine pixel-level semantic annotation, a corresponding image of the same scene under normal conditions, and a binary mask to distinguish clear and uncertain semantic content within the image. This dataset supports both standard semantic segmentation and uncertainty-aware semantic segmentation, posing challenges to state-of-the-art supervised and unsupervised approaches while guiding future progress in the field.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Qin Wang et al.
Summary: Domain adaptation for semantic segmentation can be improved by leveraging self-supervised depth estimation guidance from both source and target domains. This approach involves learning task feature correlation and utilizing depth prediction discrepancies between source and target to refine semantic segmentation. The proposed method demonstrates state-of-the-art performance on benchmark tasks SYNTHIA-to-Cityscapes and GTA-to-Cityscapes.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Shiqi Yang et al.
Summary: G-SFDA is a new domain adaptation paradigm that aims to perform well on both target and source domains, with only access to current unlabeled target data during adaptation. The proposed LSC and SDA methods successfully adapt the model to the target domain while keeping source performance.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Yuang Liu et al.
Summary: SFDA is a source-free domain adaptation framework for semantic segmentation, which recovers and preserves source domain knowledge through knowledge transfer, extracts valuable information from the target domain for self-supervised learning, and seamlessly integrates pixel- and patch-level optimization objectives tailored for semantic segmentation in the framework.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Riccardo Volpi et al.
Summary: Standard learning approaches often result in fragile models that are prone to catastrophic forgetting, but we propose a domain randomization technique to build more robust models. By randomizing the distribution of the current domain, our approach shows reduced vulnerability to catastrophic forgetting when transferred to new domains through meta-learning.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Boyi Li et al.
Summary: In this paper, the authors propose Moment Exchange, an implicit data augmentation method that encourages models to utilize moment information for both training and image recognition. By replacing and interpolating moments of training images, the model is forced to extract training signals from moments in addition to normalized features. The approach operates entirely in feature space, is fast, and can be effectively combined with existing augmentation methods to improve the generalization capability of competitive baseline networks consistently.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Minhao Hu et al.
Summary: In this paper, a novel fully test-time unsupervised adaptation method for image segmentation using Regional Nuclear-norm (RN) and Contour Regularization (CR) is proposed. By modifying parameters in batch normalization layers instead of retraining all parameters, the method efficiently improves performance and efficiency in segmentation tasks.
MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2021, PT III
(2021)
Review
Computer Science, Artificial Intelligence
German I. Parisi et al.
Article
Computer Science, Theory & Methods
Connor Shorten et al.
JOURNAL OF BIG DATA
(2019)
Article
Computer Science, Artificial Intelligence
Zhizhong Li et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2018)
Article
Computer Science, Hardware & Architecture
Alex Krizhevsky et al.
COMMUNICATIONS OF THE ACM
(2017)
Article
Engineering, Electrical & Electronic
Vishal M. Patel et al.
IEEE SIGNAL PROCESSING MAGAZINE
(2015)
Proceedings Paper
Computer Science, Artificial Intelligence
Judy Hoffman et al.
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2014)
Article
Computer Science, Artificial Intelligence
Sinno Jialin Pan et al.
IEEE TRANSACTIONS ON NEURAL NETWORKS
(2011)
