检索结果-内蒙古大学图书馆

9th International Conference on analysis of Images, Social networks, and Texts, AIST 2020

作者： Gavrikov, Ilya Savchenko, Andrey V. HSE University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (纸本)9783030712136

In this paper we study the problem of predicting the cohesiveness and emotion of a group of people in photo. We proposed a fast approach, consisting of face detection by using MTCNN, aggregation of facial features (age, gender and embeddings) extracted by multi-task MobileNet, prediction of group cohesion and classification of emotional background using multi-output convolution neural network. Experimental study on the Group Affect Dataset from EmotiW 2019 challenge demonstrated that our approach allows to achieve an improvement of quality and even to reduce the running time of an algorithm’s work when compared to known solutions. As a result, we obtained mean squared error 0.63 for cohesion prediction, which is 0.21 lower when compared to baseline CapsNet. © 2021, Springer Nature Switzerland AG.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Group-Level Affect Recognition in Video Using Deviation of Frame Features 10th

Group-Level Affect Recognition in Video Using Deviation of...

引用

10th International Conference on analysis of Images, Social networks and Texts, AIST 2021

作者： Savchenko, Andrey V. Savchenko, Lyudmila V. Belova, Natalya S. HSE University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia HSE University Moscow Russia

ISBN: (纸本)9783031164996

In this paper, we propose the novel video-based group-level emotion recognition algorithm. At first, the faces are detected in each video frame, and their features are extracted using a lightweight neural network, e.g., MobileNet pre-trained on large emotional dataset, such as AffectNet. The frame descriptor is defined as a concatenation of STAT features (max, average, standard deviation, etc.). The descriptor of the whole video is computed as a deviation of the frame descriptors, and the resulting video features are fed into a classifier. Experimental results for the VGAF dataset from the EmotiW 2020 challenge demonstrate that the proposed approach has 1% greater accuracy than the best-known single model. It is also at least 5% better than any other facial processing technique. Moreover, a blending of facial expression recognition with a processing of audio features extracted by the OpenSMILE library is comparable with the best-known ensemble for this dataset. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Neural network model for video-based facial expression recognition in-the-wild on mobile devices 7

Neural network model for video-based facial expression recog...

引用

7th IEEE International Conference on Information Technology and Nanotechnology, ITNT 2021

作者： Demochkina, Polina Savchenko, Andrey V. HSE University Nizhny Novgorod Russia HSE University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (纸本)9781665432177

In this paper, we propose to solve the problem of facial expression recognition in videos by implementing a two-stage procedure, in which, firstly, facial features are extracted from all frames using an EfficientNet-based model. The latter is pre-trained to identify facial attributes and further fine-tuned on an external dataset for the emotion classification task. Secondly, multiple statistical functions are calculated and used in the aggregation process to create a single video representation. Furthermore, we propose a new technique for sequence, frame-level attention models, and 1D convolutions by concatenating the output of a statistical function with the facial features. It was experimentally shown that the proposed approach leads to state-of-the-art results on the AFEW 8.0 dataset. © 2021 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Summarization of Videos from Online Events Based on Multimodal Emotion Recognition

Summarization of Videos from Online Events Based on Multimod...

引用

2022 International Russian Automation Conference, RusAutoCon 2022

作者： Abdrahimov, Amir Savchenko, Andrey V. Hse University Nizhny Novgorod Russia Hse University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (数字)9781665466554

ISBN: (纸本)9781665466554

In this paper, we propose a novel video summarization technique for automatic affect analysis of participants of an online event. At first, face verification neural network is used to cluster facial regions that correspond to each participant. Next, emotional features are extracted from each face by using EfficientNet model obtained in the previous paper of the author. The features of several consecutive frames are combined into a single descriptor that is used to classify emotions. In addition, audio features are extracted using wav2vec, and an ensemble of audio and video classifiers predicts emotions for each face. Finally, dependence of these emotions on time is visualized in special color charts. In the experimental study with the AFEW dataset it was demonstrated that the proposed approach makes it possible to obtain the best-known validation accuracy 67.88%. The models were optimized using OpenVINO and gained reasonable performance even if Nvidia GPUs are unavailable. The models and source code are publicly available at https://***/amirabdrahimov/multimodal-emotion-recognition. © 2022 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Touching the Limits of a Dataset in Video-Based Facial Expression Recognition

Touching the Limits of a Dataset in Video-Based Facial Expre...

引用

2021 International Russian Automation Conference, RusAutoCon 2021

作者： Churaev, Egor Savchenko, Andrey V. Hse University Nizhny Novgorod Russia Hse University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (纸本)9780738131566

In this paper, we examine the issue of video-based facial emotion recognition algorithms which show excellent performance on some benchmarks, but have much worse accuracy in practical applications. For example, the typical error rate of contemporary deep neural networks on the RAVDESS dataset is less than 5%. We argue that such results are obtained only if the split of the whole dataset is incorrect, so that the same persons are present in both training and test sets. It is claimed that it is more frankly to use the actor-based split, in which persons in the training and test sets are disjoint. It is experimentally demonstrated that the near state-of-the-art neural network model pre-trained on the AffectNet dataset achieves 99% accuracy on conventional split of the RAVDESS dataset. However, when we split the dataset by the actors and training and testing sets have only unique persons then the accuracy will be 20-30% lower. © 2021 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Deep neural networks and maximum likelihood search for approximate nearest neighbor in video-based image recognition

引用

Optical Memory and Neural networks (Information Optics) 2017年第2期26卷 129-136页

作者： Savchenko, A.V. National Research University Higher School of Economics Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

We analyzed the way to increase computational efficiency of video-based image recognition methods with matching of high dimensional feature vectors extracted by deep convolutional neural networks. We proposed an algorithm for approximate nearest neighbor search. At the first step, for a given video frame the algorithm verifies a reference image obtained when recognizing the previous frame. After that the frame is compared with a few number of reference images. Each next examined reference image is chosen so that to maximize conditional probability density of distances to the reference instances tested at previous steps. To decrease the required memory space we beforehand calculate only distances from all the images to small number of instances (pivots). When experimenting with either face photos from Labeled Faces in the Wild and PubFig83 datasets or with video data from YouTube Faces we showed that our algorithm allows accelerating the recognition procedure by 1.4–4 times comparing with known approximate nearest neighbor methods. © 2017, Allerton Press, Inc.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

EmotiEffNets for Facial Processing in Video-based Valence-Arousal Prediction, Expression Classification and Action Unit Detection

EmotiEffNets for Facial Processing in Video-based Valence-Ar...

引用

2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2023

作者： Savchenko, Andrey V. Sber Ai Lab Moscow Russia Hse University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (纸本)9798350302493

In this article, the pre-trained convolutional networks from the EmotiEffNet family for frame-level feature extraction are used for downstream emotion analysis tasks from the fifth Affective Behavior analysis in-the-wild (ABAW) competition. In particular, we propose an ensemble of a multi-layered perceptron and the LightAutoML-based classifier. The post-processing by smoothing the results for sequential frames is implemented. Experimental results for the large-scale Aff-Wild2 database demonstrate that our model is much better than the baseline facial processing using VGGFace And ResNet. For example, our macro-averaged F1-scores of facial expression recognition and action unit detection on the testing set are 11-13% greater. Moreover, the concordance correlation coefficients for valence/arousal estimation are up to 30% higher when compared to the baseline. © 2023 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

MT-EmotiEffNet for Multi-task Human Affective Behavior analysis and Learning from Synthetic Data 17th

MT-EmotiEffNet for Multi-task Human Affective Behavior Anal...

引用

17th European Conference on Computer Vision, ECCV 2022

作者： Savchenko, Andrey V. HSE University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia Sber AI Lab Moscow Russia

ISBN: (纸本)9783031250743

In this paper, we present the novel multi-task EfficientNet model and its usage in the 4th competition on Affective Behavior analysis in-the-wild (ABAW). This model is trained for simultaneous recognition of facial expressions and prediction of valence and arousal on static photos. The resulting MT-EmotiEffNet extracts visual features that are fed into simple feed-forward neural networks in the multi-task learning challenge. We obtain performance measure 1.3 on the validation set, which is significantly greater when compared to either performance of baseline (0.3) or existing models that are trained only on the s-Aff-Wild2 database. In the learning from synthetic data challenge, the quality of the original synthetic training set is increased by using the super-resolution techniques, such as Real-ESRGAN. Next, the MT-EmotiEffNet is fine-tuned on the new training set. The final prediction is a simple blending ensemble of pre-trained and fine-tuned MT-EmotiEffNets. Our average validation F1 score is 18% greater than the baseline model. As a result, our team took the first place in the learning from synthetic data task and the third place in the multi-task learning challenge. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Blending

来源：评论

学校读者我要写书评

暂无评论

Effective face recognition based on anomaly image detection and sequential analysis of neural descriptors 9

Effective face recognition based on anomaly image detection ...

引用

9th IEEE International Conference on Information Technology and Nanotechnology, ITNT 2023

作者： Sokolova, Anastasiia Savchenko, Andrey V. Hse University Nizhny Novgorod Russia Sber Ai Lab Moscow Russia Hse University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia

ISBN: (纸本)9798350397338

In this paper, we explore the possibility to improve efficiency of face recognition using information about anomaly input images. Indeed, modern publicly-available datasets typically contain images of mostly middle-aged and Caucasian people, which cause most algorithms to fail on photos of older people or children, rarer ethnicities, poor-quality images, etc. Detection of such anomaly data and its subsequent rejection helps to improve the classification accuracy. We propose a novel algorithm, at the first stage of which a convolutional neural network is used to detect anomalies in input images. This network is trained on a specially created set of rare data. The second stage is the sequential analysis of neural descriptors extracted from input faces to improve the computational efficiency of classification. An experimental study on the VGGFace2 and MS-Celeb-1M datasets using neural network descriptors, including contemporary InsightFace models, demonstrated the effectiveness of the proposed algorithm. © 2023 IEEE.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Robustness of Graphical Lasso Optimization Algorithm for Learning a Graphical Model 23rd

Robustness of Graphical Lasso Optimization Algorithm for L...

引用

23rd International Conference on Mathematical Optimization Theory and Operations Research, MOTOR 2024

作者： Kalyagin, Valeriy Kostylev, Ilya HSE University Laboratory of Algorithms and Technologies for Network Analysis Nizhny Novgorod Russia Steklov Mathematical Institute Russian Academy of Sciences Moscow Russia

ISBN: (纸本)9783031627910

Problem of learning a graphical model (graphical model selection problem) consists of recovering a conditional dependence structure (concentration graph) from data given as a sample of observations from a random vector. Various algorithms to solve this problem are known. One class of algorithms is related with convex optimization problem with additional lasso regularization term. Such algorithms are called graphical lasso algorithms. Various properties and practical efficiency of graphical lasso algorithms were investigated in the literature. In the present paper we study sensitivity of uncertainty (level of error) of graphical lasso algorithms to the change of distribution of the random vector. This issue is not well studied yet. First, we show that uncertainty of the classical version of graphical lasso algorithm is very sensitive to the change of distribution. Next, we suggest simple modifications of this algorithm which are much more robust in the large class of distributions. Finally, we discuss a future development of the proposed approach. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： Graphic methods

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：