Filters








3,961 Hits in 5.7 sec

Hard-Mining Loss based Convolutional Neural Network for Face Recognition [article]

Yash Srivastava and Vaishnav Murali and Shiv Ram Dubey
2020 arXiv   pre-print
Witnessing advances in deep learning, significant work has been observed in face recognition, which touched upon various parts of the recognition framework like Convolutional Neural Network (CNN), Layers  ...  Various loss functions such as Cross-Entropy, Angular-Softmax and ArcFace have been introduced to learn the weights of network for face recognition.  ...  The stochastic gradient descent (SGD) optimization is widely adapted to train the Convolutional Neural Networks (CNNs).  ... 
arXiv:1908.09747v2 fatcat:hpbudqcjcvfvrfiggt6sdczztu

Doppelganger Mining for Face Representation Learning

Evgeny Smirnov, Aleksandr Melnikov, Sergey Novoselov, Eugene Luckyanets, Galina Lavrentyeva
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
It is especially useful for methods, based on exemplar-based supervision.  ...  Usually hard example mining comes with a price of necessity to use large mini-batches or substantial extra computation and memory cost, particularly for datasets with large numbers of identities.  ...  In this paper we propose Doppelganger mining -a new simple sampling method, which improves the training of Deep Convolutional Neural Networks for face recognition.  ... 
doi:10.1109/iccvw.2017.226 dblp:conf/iccvw/SmirnovMNLL17 fatcat:wjo6lx6nbbbgrkn44gbypvexp4

Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks

Kaipeng Zhang, Zhanpeng Zhang, Zhifeng Li, Yu Qiao
2016 IEEE Signal Processing Letters  
In particular, our framework adopts a cascaded structure with three stages of carefully designed deep convolutional networks that predict face and landmark location in a coarse-to-fine manner.  ...  Our method achieves superior accuracy over the state-of-the-art techniques on the challenging FDDB and WIDER FACE benchmark for face detection, and AFLW benchmark for face alignment, while keeps real time  ...  [11] train deep convolution neural networks for facial attribute recognition to obtain high response in face regions which further yield candidate windows of faces.  ... 
doi:10.1109/lsp.2016.2603342 fatcat:o73anfq5sngqbgdk3zj5aowuam

Hard Example Mining with Auxiliary Embeddings

Evgeny Smirnov, Elizaveta Ivanova, Aleksandr Melnikov, Ilya Kalinovskiy, Andrei Oleinik, Eugene Luckyanets
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
Our experiments on the challenging Disguised Faces in the Wild (DFW) dataset show that hard example mining with auxiliary embeddings improves the discriminative power of learned representations.  ...  With the help of these embeddings it is possible to select new examples for the mini-batch based on their similarity with the already selected examples.  ...  In the context of face recognition, it can be achieved by means of a pre-trained attribute classification neural network.  ... 
doi:10.1109/cvprw.2018.00013 dblp:conf/cvpr/SmirnovMOIKL18 fatcat:h3jx2l6ghfccxpjmp6vlclpnju

Metric Classification Network in Actual Face Recognition Scene [article]

Jian Li, Yan Wang, Xiubao Zhang, Weihong Deng, Haifeng Shen
2019 arXiv   pre-print
These experiments confirm the effectiveness of validation classifier on face recognition task.  ...  This is very inappropriate for application in real-world scenarios.  ...  , followed by a similarity measure also based on neural network.  ... 
arXiv:1910.11563v1 fatcat:l7hmwhs2mjgtpia54bckljc2nu

MassFace: an efficient implementation using triplet loss for face recognition [article]

Yule Li
2019 arXiv   pre-print
In this paper we present an efficient implementation using triplet loss for face recognition. We conduct the practical experiment to analyze the factors that influence the training of triplet loss.  ...  We analyze the experiment results and give some insights to help others balance the factors when they apply triplet loss to their own problem especially for face recognition task.  ...  Introduction Face recognition has achieved significant improvement due to the power of deep representation through convolutional neural network.  ... 
arXiv:1902.11007v1 fatcat:dcyhgwmttzgjle6m5raf2jcl3i

Detection of Pin Defects in Aerial Images Based on Cascaded Convolutional Neural Network

Yewei Xiao, Zhiqiang Li, Dongbo Zhang, Lianwei Teng
2021 IEEE Access  
INDEX TERM pin defect, aerial image, cascaded convolutional neural network, nonlinear multilayer perceptron, hard sample mining  ...  This paper proposed a target detection method based on cascaded convolutional neural networks.  ...  For pre-training, we adopt the strategy of online hard sample mining (OHEM) [23] .  ... 
doi:10.1109/access.2021.3079172 fatcat:ktgmcvpb75exhcmkomvhy447by

Real-Time Face Recognition System for Remote Employee Tracking [article]

Mohammad Sabik Irbaz, MD Abdullah Al Nasim, Refat E Ferdous
2021 arXiv   pre-print
To deal with the challenge effectively, we came up with a solution to track the employees with face recognition. We have been testing this system experimentally for our office.  ...  To train the face recognition module, we used FaceNet with KNN using the Labeled Faces in the Wild (LFW) dataset and achieved 97.8\% accuracy.  ...  For identifying faces two approaches are explained here. First one is effective Convolutional Neural Network designs for biometric face recognition.  ... 
arXiv:2107.07576v2 fatcat:xnmwlzofpbbt7f4uanqcuuzpia

Improved YOLOv3 Object Classification in Intelligent Transportation System [article]

Yang Zhang, Changhui Hu, Xiaobo Lu
2020 arXiv   pre-print
The proposed model and contrast experiment are conducted on our self-build traffic driver's face database.  ...  In this paper, an algorithm based on YOLOv3 is proposed to realize the detection and classification of vehicles, drivers, and people on the highway, so as to achieve the purpose of distinguishing driver  ...  [7] proposed the deep neural networks for facial feature recognition, aiming at getting high reply in face regions, resulting in producing candidate windows of human faces.  ... 
arXiv:2004.03948v1 fatcat:fr2o5wafvjdhlofdszelcje37i

LANDMARK-BASED VISUAL PLACE RECOGNITION

Swapnali Gavali, Dr. Bashirahamad Momin
2021 International Journal of Engineering Applied Sciences and Technology  
By fine-tuning pre-trained convolutional neural network (CNN) and minimizing triplet loss, the triplet network can learn appropriate metrics so that most similar images can be retrieved through algorithms  ...  This paper presents the application of a triplet network for large scale landmarkbased visual place recognition.  ...  [7] In this study, examined Siamese convolutional neural network architectures to verify authorship of handwritten text.  ... 
doi:10.33564/ijeast.2021.v05i11.035 fatcat:p7dwn5r4ivhcdko7n2nuin6fya

Bootstrapping Face Detection with Hard Negative Examples [article]

Shaohua Wan, Zhijun Chen, Tao Zhang, Bo Zhang, Kong-kat Wong
2016 arXiv   pre-print
The key is to exploit the idea of hard negative mining and iteratively update the Faster R-CNN based face detector with the hard negatives harvested from a large set of background examples.  ...  Recently significant performance improvement in face detection was made possible by deeply trained convolutional networks.  ...  Since the remarkable success of the deep Convolutional Neural Network (CNN) [10] in image classification on the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2012, numerous efforts have  ... 
arXiv:1608.02236v1 fatcat:56wkjnzbdnehpfv52pvg3hjujq

TinaFace: Strong but Simple Baseline for Face Detection [article]

Yanjia Zhu, Hongxiang Cai, Shuhan Zhang, Chenhao Wang, Yichao Xiong
2021 arXiv   pre-print
On the hard test set of the most popular and challenging face detection benchmark WIDER FACE , with single-model and single-scale, our TinaFace achieves 92.1% average precision (AP), which exceeds most  ...  Many works present lots of special methods for face detection from different perspectives like model architecture, data augmentation, label assignment and etc., which make the overall algorithm and system  ...  “Face detection through scale- over union: A metric and a loss for bounding box friendly deep convolutional networks”. In: arXiv regression”.  ... 
arXiv:2011.13183v3 fatcat:5vqpkodpgzbpplj5jyog3pr6nu

Are Gabor Kernels Optimal for Iris Recognition? [article]

Aidan Boyd, Adam Czajka, Kevin Bowyer
2020 arXiv   pre-print
We use (on purpose) a single-layer convolutional neural network as it mimics an iris code-based algorithm.  ...  Gabor kernels are widely accepted as dominant filters for iris recognition.  ...  We propose to use a single-layer convolutional neural network that replicates a Daugman's approach to iris recognition.  ... 
arXiv:2002.08959v1 fatcat:frrssua4zzbdlh7kbbpr5ppzgu

DenseBox: Unifying Landmark Localization with End to End Object Detection [article]

Lichao Huang and Yi Yang and Yafeng Deng and Yinan Yu
2015 arXiv   pre-print
How can a single fully convolutional neural network (FCN) perform on object detection?  ...  We present experimental results on public benchmark datasets including MALF face detection and KITTI car detection, that indicate our DenseBox is the state-of-the-art system for detecting challenging objects  ...  [10] , is a face detection system based on convolutional neural networks.  ... 
arXiv:1509.04874v3 fatcat:ycaoxizg4bbqxagyvueieb6vpu

Seeing What is Not There: Learning Context to Determine Where Objects are Missing

Jin Sun, David W. Jacobs
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Our model is based on a convolutional neural network structure. With a specially designed training strategy, the model learns to ignore objects and focus on context only.  ...  It is fully convolutional thus highly efficient.  ...  Figure 9 : 9 Retrieved out of context faces by a SFC network. Table 2 : 2 Neural network structure summary for the base network.  ... 
doi:10.1109/cvpr.2017.136 dblp:conf/cvpr/SunJ17 fatcat:lfaic5wrp5gard7qxtchcd3ohi
« Previous Showing results 1 — 15 out of 3,961 results