Filters








12 Hits in 3.2 sec

Learning a Metric Embedding for Face Recognition using the Multibatch Method [article]

Oren Tadmor and Yonatan Wexler and Tal Rosenwein and Shai Shalev-Shwartz and Amnon Shashua
2016 arXiv   pre-print
Our main technical contribution centers around a novel training method, called Multibatch, for similarity learning, i.e., for the task of generating an invariant "face signature" through training pairs  ...  This work is motivated by the engineering task of achieving a near state-of-the-art face recognition on a minimal computing budget running on an embedded system.  ...  We have shown that our multibatch method introduces a dramatic acceleration of the training time required to learn an effective embedding for the task of face recognition.  ... 
arXiv:1605.07270v1 fatcat:lawhovo76jc6dac2uztpprrvqq

Marginal Loss for Deep Face Recognition

Jiankang Deng, Yuxiang Zhou, Stefanos Zafeiriou
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
In order to enhance the discriminative power of the deeply learned features, we propose a new supervision signal named marginal loss for deep face recognition.  ...  Convolutional neural networks have significantly boosted the performance of face recognition in recent years due to its high capacity in learning discriminative features.  ...  Introduction Face representation through the deep convolutional network embedding is considered the state-of-the-art method for face verification, face clustering, and recognition [24, 18, 17] .  ... 
doi:10.1109/cvprw.2017.251 dblp:conf/cvpr/DengZZ17 fatcat:j46bq6jlmjetpkmsiw5clisxgq

Joint Multi-Patch and Multi-Task CNNs for Robust Face Recognition

Yanfei LIU, Junhua CHEN, Yu QIU
2020 IEICE transactions on information and systems  
recognition task and pose estimation task, is operated on the fused feature to obtain a pose-invariant face representation for the face recognition task.  ...  In this paper, we present a joint multi-patch and multitask convolutional neural networks (JMM-CNNs) framework to learn more descriptive and robust face representation for face recognition.  ...  The proposed method falls under the multi-patch CNN method and the multi-task learning approach with CNNs for face recognition, but with several differences compared with existing methods.  ... 
doi:10.1587/transinf.2020edp7059 fatcat:i4g4ndbjhfczrerto5osakdone

Recent Advances in Deep Learning Techniques for Face Recognition [article]

Md. Tahmid Hasan Fuad, Awal Ahmed Fime, Delowar Sikder, Md. Akil Raihan Iftee, Jakaria Rabbi, Mabrook S. Al-rakhami, Abdu Gumae, Ovishake Sen, Mohtasim Fuad, Md. Nazrul Islam
2021 arXiv   pre-print
In recent years, researchers have proposed many deep learning (DL) methods for various tasks, and particularly face recognition (FR) made an enormous leap using these techniques.  ...  Deep FR systems benefit from the hierarchical architecture of the DL methods to learn discriminative face representation.  ...  Multibatch [126] Metric learning 2.6M 98.20 Reduce computational cost and increase precision.  ... 
arXiv:2103.10492v1 fatcat:h526swzntjgmlcjmwnuidqg44u

Deep Metric Learning by Online Soft Mining and Class-Aware Attention

Xinshao Wang, Yang Hua, Elyor Kodirov, Guosheng Hu, Neil M. Robertson
2019 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Deep metric learning aims to learn a deep embedding that can capture the semantic similarity of data points.  ...  Given the availability of massive training samples, deep metric learning is known to suffer from slow convergence due to a large fraction of trivial samples.  ...  The goal of deep metric learning is to learn a feature embedding/representation that captures the semantic similarity of data points in the embedding space such that samples belonging to the same classes  ... 
doi:10.1609/aaai.v33i01.33015361 fatcat:wklvfundbjd25lb2pthegegzam

Rethinking Feature Discrimination and Polymerization for Large-scale Recognition [article]

Yu Liu and Hongyang Li and Xiaogang Wang
2017 arXiv   pre-print
It inherits the softmax property to make inter-class features discriminative as well as shares the idea of class centroid in metric learning.  ...  Unlike previous work where the center is a temporal, statistical variable within one mini-batch during training, the formulated centroid is responsible for clustering inner-class features to enforce them  ...  The feature dimension is three for better illustration. The triplet loss [33] has a constraint on feature embedding ( f 2 = 1) and thus the learned features are scatted on a hypersphere.  ... 
arXiv:1710.00870v2 fatcat:qsu3waacuzbdjolani35jarrjm

Deep Metric Learning by Online Soft Mining and Class-Aware Attention [article]

Xinshao Wang, Yang Hua, Elyor Kodirov, Guosheng Hu, Neil M. Robertson
2019 arXiv   pre-print
Deep metric learning aims to learn a deep embedding that can capture the semantic similarity of data points.  ...  Given the availability of massive training samples, deep metric learning is known to suffer from slow convergence due to a large fraction of trivial samples.  ...  The cropped images are used for training and testing for methods in the second group. * indicates cascaded models are applied for sample mining and learning embeddings.  ... 
arXiv:1811.01459v3 fatcat:ntm2v4vjcfdfhmlhhjjkle7y7u

Deep Face Recognition: A Survey [article]

Mei Wang, Weihong Deng
2020 arXiv   pre-print
Then, we summarize and compare the commonly used databases for both model training and evaluation.  ...  This emerging technique has reshaped the research landscape of face recognition (FR) since 2014, launched by the breakthroughs of DeepFace and DeepID.  ...  [232] proposed a hard example mining method benefitted from class-wise (Doppelganger Mining [233] ) and example-wise mining to learn useful deep embeddings for disguised face recognition.  ... 
arXiv:1804.06655v9 fatcat:i2yxh7bf45attlfjd3akyc65me

Generalization in Metric Learning: Should the Embedding Layer be the Embedding Layer? [article]

Nam Vo, James Hays
2018 arXiv   pre-print
In particular, we investigate using other layers in a deep metric learning system (besides the embedding layer) for feature extraction and analyze how well they perform on training data and generalize  ...  This work studies deep metric learning under small to medium scale data as we believe that better generalization could be a contributing factor to the improvement of previous fine-grained image retrieval  ...  4, 8] and face recognition [26, 19] .  ... 
arXiv:1803.03310v2 fatcat:5u5bdzp2zzco3ax3gffwvfdf34

Egocentric video summarisation via purpose-oriented frame scoring and selection

V. Javier Traver, Dima Damen
2021 Expert systems with applications  
On the one hand, a partially agnostic method uses the scores obtained by the proposed approach, but follows a standard generic frame selection technique.  ...  On the other hand, the fully agnostic method do not use any purposebased information, and relies on generic concepts such as diversity and representativeness.  ...  The financial support for the research stay with code PRX18/00283, and for the research network with code RED2018-102511-T, both from Ministerio de Ciencia, Innovación y Universidades, are acknowledged  ... 
doi:10.1016/j.eswa.2021.116079 fatcat:k7uoocuarvasrnqjjbl6chgaie

A framework for operative and social sustainability functionalities in Human-Centric Cyber-Physical Production Systems

Marta Pinzone, Federico Albè, Davide Orlandelli, Ilaria Barletta, Cecilia Berlin, Björn Johansson, Marco Taisch
2018 Computers & industrial engineering  
In a near future where manufacturing companies are faced with the rapid technological developments of Cyber-Physical Systems (CPS) and Industry 4.0, a need arises to consider how this will affect human  ...  Finally, it presents an industrial use case, which the CyFL Matrix and the related guidelines are applied to.  ...  Acknowledgements The Authors would like to thank Whirlpool EMEA for participating in the research.  ... 
doi:10.1016/j.cie.2018.03.028 fatcat:4tq64na6tvdnbahfislhzw5xaq

Studying Myeloid Cell Heterogeneity After Spinal Cord Injury via Time-Resolved Single-Cell RNA Sequencing

Regan Hamel, Apollo-University Of Cambridge Repository, Stefano Pluchino, John Marioni
2022
Spinal cord injury (SCI) is a devastating pathology that affects thousands of individuals annually, resulting in the requirement for long-term physical and medical care and thus significant personal, societal  ...  To fully appreciate the temporal dynamics of the pathology, I collected samples across the acute, subacute, and early chronic phases of SCI, plus a sham-injured control.  ...  ACKNOWLEDGEMENTS Thanks to Stefano Pluchino for his supervision. Bryan, Veronica, and Yutong, thank you all for  ... 
doi:10.17863/cam.81960 fatcat:ubdfjsynvje7dffbthdvv2roaq