Filters








11,177 Hits in 6.4 sec

Label-Assemble: Leveraging Multiple Datasets with Partial Labels [article]

Mintong Kang, Yongyi Lu, Alan L. Yuille, Zongwei Zhou
2022 arXiv   pre-print
From rigorous evaluations on three natural imaging and six medical imaging tasks, we discover that learning from "negative examples" facilitates both classification and segmentation of classes of interest  ...  Remarkably, when using existing partial labels, our model performance is on-par with that using full labels, eliminating the need for an additional 40% of annotation costs.  ...  We also thank the constructive suggestions from Dr. J. Liang, Dr. Z. Zhu, L. Chen, J. Chen, Y. Bai, G. Li, and X. Li.  ... 
arXiv:2109.12265v3 fatcat:o6vg4gmlazbxdmvvqyiuq6oliy

mDALU: Multi-Source Domain Adaptation and Label Unification with Partial Datasets [article]

Rui Gong, Dengxin Dai, Yuhua Chen, Wen Li, Luc Van Gool
2021 arXiv   pre-print
Extensive experiments on three different tasks - image classification, 2D semantic image segmentation, and joint 2D-3D semantic segmentation - show that our method outperforms all competing methods significantly  ...  In the former, partial knowledge is transferred from multiple source domains to the target domain and fused therein.  ...  This research has received funding from the EU Horizon 2020 research and innovation programme under grant agreement No. 820434. Dengxin Dai is the corresponding author.  ... 
arXiv:2012.08385v2 fatcat:mov5x43bjjbgljzkcalkhyaxt4

Label Propagation for Annotation-Efficient Nuclei Segmentation from Pathology Images [article]

Yi Lin, Zhiyong Qu, Hao Chen, Zhongke Gao, Yuexiang Li, Lili Xia, Kai Ma, Yefeng Zheng, Kwang-Ting Cheng
2022 arXiv   pre-print
First, coarse pixel-level labels are derived from the point annotations based on the Voronoi diagram and the k-means clustering method to avoid overfitting.  ...  Generally, the segmentation performance of fully-supervised learning heavily depends on the amount and quality of the annotated data.  ...  We investigate three data-split strategies: completely non-overlapping, partially overlapping, and completely overlapping.  ... 
arXiv:2202.08195v1 fatcat:3nbez677xrgohlgt5odagv6vv4

Federated Multi-organ Segmentation with Partially Labeled Data [article]

Xuanang Xu, Pingkun Yan
2022 arXiv   pre-print
In practice, each clinical site may only annotate certain organs of interest with partial or no overlap with other sites.  ...  Federated learning is an emerging paradigm allowing large-scale decentralized learning without sharing data across different data owners, which helps address the concern of data privacy in medical image  ...  Medical image segmentation with partial labels Due to the high cost of data annotation, medical images are often partially labeled with different ROIs or labels, even though they may share the same imaging  ... 
arXiv:2206.07156v1 fatcat:7tkehuaikzebtjwwx24bj35jha

Multi-domain semantic segmentation with overlapping labels [article]

Petra Bevandić, Marin Oršić, Ivan Grubišić, Josip Šarić, Siniša Šegvić
2021 arXiv   pre-print
We address this challenge by proposing a principled method for seamless learning on datasets with overlapping classes based on partial labels and probabilistic loss.  ...  Our method achieves competitive within-dataset and cross-dataset generalization, as well as ability to learn visual concepts which are not separately labeled in any of the training datasets.  ...  Learning from partial labels considers examples labeled with a set of classes only one of which is correct.  ... 
arXiv:2108.11224v2 fatcat:qqxnvmcckbftbmonvaf4qx5cpq

Universal Lesion Detection by Learning from Multiple Heterogeneously Labeled Datasets [article]

Ke Yan, Jinzheng Cai, Adam P. Harrison, Dakai Jin, Jing Xiao, Le Lu
2020 arXiv   pre-print
In this way, reliable positive and negative regions are obtained from partially-labeled and unlabeled images, which are effectively utilized to train ULD.  ...  Lesion detection is an important problem within medical imaging analysis. Most previous work focuses on detecting and segmenting a specialized category of lesions (e.g., lung nodules).  ...  missing annotations Negative region mining Single-type 1 proposals Single-type 2 proposals Existing annotations Matched missing annotations The rest part of the image Partially- labeled  ... 
arXiv:2005.13753v1 fatcat:cpsbd73ew5baxp2slt2hipha6q

Learning from Multiple Datasets with Heterogeneous and Partial Labels for Universal Lesion Detection in CT [article]

Ke Yan, Jinzheng Cai, Youjing Zheng, Adam P. Harrison, Dakai Jin, Youbao Tang, Yuxing Tang, Lingyun Huang, Jing Xiao, Le Lu
2021 arXiv   pre-print
However, due to the annotation cost, datasets in medical imaging are often either partially-labeled or small.  ...  Next, we propose strategies to mine missing annotations from partially-labeled datasets by exploiting clinical prior knowledge and cross-dataset knowledge transfer.  ...  For each test image, we generate proposals from all dataset experts, then do NMS to filter the overlapped boxes.  ... 
arXiv:2009.02577v3 fatcat:pfw4h3yq5ndtnctj26ufuak3sa

Adding Seemingly Uninformative Labels Helps in Low Data Regimes [article]

Christos Matsoukas, Albert Bou I Hernandez, Yue Liu, Karin Dembrower, Gisele Miranda, Emir Konuk, Johan Fredin Haslum, Athanasios Zouzos, Peter Lindholm, Fredrik Strand, Kevin Smith
2020 arXiv   pre-print
In this work, we consider a task that requires difficult-to-obtain expert annotations: tumor segmentation in mammography images.  ...  We show that, in low-data settings, performance can be improved by complementing the expert annotations with seemingly uninformative labels from non-expert annotators, turning the task into a multi-class  ...  Acknowledgements This work was partially supported by the Wallenberg Autonomous Systems Program (WASP), the Swedish Research Council (VR) 2017-04609, and Region Stockholm HMT 20170802.  ... 
arXiv:2008.00807v2 fatcat:t5tvbbq7tzbjlo46jhjtp2c5bq

Factorisation-based Image Labelling [article]

Yu Yan, Yael Balbastre, Mikael Brudfors, John Ashburner
2021 arXiv   pre-print
Manual annotation is time consuming and expensive, so having a fully automated and general purpose brain segmentation algorithm is highly desirable.  ...  As our approach is intended to be general purpose, we also assess how well it can handle domain shift by labelling images of the same subjects acquired with different MR contrasts.  ...  Similarly, the model could be learned by combining sets of annotations defined using different labelling protocols.  ... 
arXiv:2111.10326v2 fatcat:jubdlnpbdrhz3ktdngb7o5nz7i

Continual Active Learning Using Pseudo-Domains for Limited Labelling Resources and Changing Acquisition Characteristics [article]

Matthias Perkonigg, Johannes Hofmanninger, Christian Herold, Helmut Prosch, Georg Langs
2022 arXiv   pre-print
The approach automatically recognizes shifts in image acquisition characteristics - new domains -, selects optimal examples for labelling and adapts training accordingly.  ...  However, continual manual expert labelling of medical imaging requires substantial effort.  ...  Acknowledgments This work was partially supported by the Austrian Science Fund (FWF): P 35189, by the Vienna Science and Technology Fund (WWTF): LS20-065, and by Novartis Pharmaceuticals Corporation.  ... 
arXiv:2111.13069v2 fatcat:jlejzfxpkbd3raxllazwppmcfe

A Survey on Label-efficient Deep Segmentation: Bridging the Gap between Weak Supervision and Dense Prediction [article]

Wei Shen, Zelin Peng, Xuehui Wang, Huayu Wang, Jiazhong Cen, Dongsheng Jiang, Lingxi Xie, Xiaokang Yang, Qi Tian
2022 arXiv   pre-print
To alleviate this burden, the past years have witnessed an increasing attention in building label-efficient, deep-learning-based segmentation algorithms.  ...  Next, we summarize the existing label-efficient segmentation methods from a unified perspective that discusses an important question: how to bridge the gap between weak supervision and dense prediction  ...  this section, we review the methods to perform instance segmentation under the partially-supervised setting.  ... 
arXiv:2207.01223v1 fatcat:i7rgpxrfkrdbfm4effjdcjjr24

Spatial role labeling

Parisa Kordjamshidi, Martijn Van Otterlo, Marie-Francine Moens
2011 ACM Transactions on Speech and Language Processing  
In Section 4, we describe our approach, based on machine learning techniques, to learn the spatial role labeling task from an annotated dataset.  ...  -We demonstrate the injection of external data resources into the spatial role labeling task by exploiting sense-annotated prepositions from TPP and compare it to a one-step approach, limited to only using  ...  Moreover, using partially labeled data in the joint learning setting would gradually decrease the requirements for manually prepared, annotated data.  ... 
doi:10.1145/2050104.2050105 dblp:journals/tslp/KordJamshidiOM11 fatcat:omawehhwafdwvnoysbggea5hii

Factorisation-Based Image Labelling

Yu Yan, Yaël Balbastre, Mikael Brudfors, John Ashburner
2022 Frontiers in Neuroscience  
Manual annotation is time consuming and expensive, so having a fully automated and general purpose brain segmentation algorithm is highly desirable.  ...  As our approach is intended to be general purpose, we also assess how well it can handle domain shift by labelling images of the same subjects acquired with different MR contrasts.  ...  Replicability Under Domain Shift This section assesses the replicability of the proposed label propagation method, by computing DSC between labellings computed from T1w scans, versus those obtained from  ... 
doi:10.3389/fnins.2021.818604 pmid:35110992 pmcid:PMC8801908 fatcat:xapxqp45lrdojazkmi3r7a2jwi

Label-Efficient Point Cloud Semantic Segmentation: An Active Learning Approach [article]

Xian Shi, Xun Xu, Ke Chen, Lile Cai, Chuan Sheng Foo, Kui Jia
2021 arXiv   pre-print
Deep learning models are the state-of-the-art methods for semantic point cloud segmentation, the success of which relies on the availability of large-scale annotated datasets.  ...  cloud segmentation.  ...  As a result, the community recently turned to exploiting partially labelled data [31] , inexactly labelled data [26] and weakly labelled data [20] for the sake of saving annotation cost.  ... 
arXiv:2101.06931v2 fatcat:awmakdxm5bawhaimd55wosgkxe

Probabilistic Joint Image Segmentation and Labeling by Figure-Ground Composition

Adrian Ion, João Carreira, Cristian Sminchisescu
2013 International Journal of Computer Vision  
image interpretations (tilings) composed from those segments, and over their labeling into categories.  ...  We propose a layered statistical model for image segmentation and labeling obtained by combining independently extracted, possibly overlapping sets of figure-ground (FG) segmentations.  ...  Acknowledgments This work was supported, in part, by CNCS-UEFICSDI, under PCE-2011-3-0438, and CT-ERC-2012-1, and by FCT under PTDC/EEA-CRO/122812/2010.  ... 
doi:10.1007/s11263-013-0663-7 fatcat:bp65aler4jhirgxv6xy2svk6lq
« Previous Showing results 1 — 15 out of 11,177 results