scholarly journals Real-Time Facial Affective Computing on Mobile Devices

Sensors ◽  
2020 ◽  
Vol 20 (3) ◽  
pp. 870 ◽  
Author(s):  
Yuanyuan Guo ◽  
Yifan Xia ◽  
Jing Wang ◽  
Hui Yu ◽  
Rung-Ching Chen
Keyword(s):  
Computational Complexity ◽  
Real Time ◽  
Mobile Devices ◽  
High Performance ◽  
Affective Computing ◽  
State Of The Art ◽  
Light Weight ◽  
Computational Power ◽  
Art Methods ◽  
And Storage

Convolutional Neural Networks (CNNs) have become one of the state-of-the-art methods for various computer vision and pattern recognition tasks including facial affective computing. Although impressive results have been obtained in facial affective computing using CNNs, the computational complexity of CNNs has also increased significantly. This means high performance hardware is typically indispensable. Most existing CNNs are thus not generalizable enough for mobile devices, where the storage, memory and computational power are limited. In this paper, we focus on the design and implementation of CNNs on mobile devices for real-time facial affective computing tasks. We propose a light-weight CNN architecture which well balances the performance and computational complexity. The experimental results show that the proposed architecture achieves high performance while retaining the low computational complexity compared with state-of-the-art methods. We demonstrate the feasibility of a CNN architecture in terms of speed, memory and storage consumption for mobile devices by implementing a real-time facial affective computing application on an actual mobile device.

scholarly journals User-independent accelerometer-based gesture recognition for mobile devices

2013 ◽  
Vol 1 (3) ◽  
pp. 11-25 ◽  
Author(s):  
Xian Wang ◽  
Paula Tarrío ◽  
Ana María Bernardos ◽  
Eduardo Metola ◽  
José Ramón Casar
Keyword(s):  
Computational Complexity ◽  
Mobile Devices ◽  
Gesture Recognition ◽  
Inertial Sensors ◽  
Recognition Accuracy ◽  
State Of The Art ◽  
Smart Phone ◽  
Recognition System ◽  
Independent Manner ◽  
Wheeled Robot

Many mobile devices embed nowadays inertial sensors. This enables new forms of human-computer interaction through the use of gestures (movements performed with the mobile device) as a way of communication. This paper presents an accelerometer-based gesture recognition system for mobile devices which is able to recognize a collection of 10 different hand gestures. The system was conceived to be light and to operate in a user-independent manner in real time. The recognition system was implemented in a smart phone and evaluated through a collection of user tests, which showed a recognition accuracy similar to other state-of-the art techniques and a lower computational complexity. The system was also used to build a human-robot interface that enables controlling a wheeled robot with the gestures made with the mobile phone

scholarly journals Soft Threshold Ternary Networks

2020 ◽  
Author(s):  
Weixiang Xu ◽  
Xiangyu He ◽  
Tianli Zhao ◽  
Qinghao Hu ◽  
Peisong Wang ◽  
...  
Keyword(s):  
Neural Networks ◽  
Mobile Devices ◽  
State Of The Art ◽  
Performance Gap ◽  
Training Time ◽  
Current State ◽  
The Arts ◽  
Soft Threshold ◽  
And Storage ◽  
Selection Of

Large neural networks are difficult to deploy on mobile devices because of intensive computation and storage. To alleviate it, we study ternarization, a balance between efficiency and accuracy that quantizes both weights and activations into ternary values. In previous ternarized neural networks, a hard threshold Δ is introduced to determine quantization intervals. Although the selection of Δ greatly affects the training results, previous works estimate Δ via an approximation or treat it as a hyper-parameter, which is suboptimal. In this paper, we present the Soft Threshold Ternary Networks (STTN), which enables the model to automatically determine quantization intervals instead of depending on a hard threshold. Concretely, we replace the original ternary kernel with the addition of two binary kernels at training time, where ternary values are determined by the combination of two corresponding binary values. At inference time, we add up the two binary kernels to obtain a single ternary kernel. Our method dramatically outperforms current state-of-the-arts, lowering the performance gap between full-precision networks and extreme low bit networks. Experiments on ImageNet with AlexNet (Top-1 55.6%), ResNet-18 (Top-1 66.2%) achieves new state-of-the-art.

A Tour of Lattice-Based Skyline Algorithms

2018 ◽  
pp. 96-122
Author(s):  
Markus Endres ◽  
Lena Rudenko
Keyword(s):  
Real Time ◽  
Data Streams ◽  
High Performance ◽  
State Of The Art ◽  
Experimental Results ◽  
Lattice Structures ◽  
Skyline Query ◽  
Basic Concepts ◽  
Generic Index

A skyline query retrieves all objects in a dataset that are not dominated by other objects according to some given criteria. There exist many skyline algorithms which can be classified into generic, index-based, and lattice-based algorithms. This chapter takes a tour through lattice-based skyline algorithms. It summarizes the basic concepts and properties, presents high-performance parallel approaches, shows how one overcomes the low-cardinality restriction of lattice structures, and finally presents an application on data streams for real-time skyline computation. Experimental results on synthetic and real datasets show that lattice-based algorithms outperform state-of-the-art skyline techniques, and additionally have a linear runtime complexity.

Numerical analysis of state of the art high performance thermoplastic composite as light weight bullet proof material

2019 ◽  
Vol 6 (9) ◽  
pp. 095333 ◽  
Author(s):  
Meenakshi S ◽  
Neeraj Srinivas ◽  
Y Sai Siddarth ◽  
Ch V S Kamal ◽  
Sudheendra K ◽  
...  
Keyword(s):  
Numerical Analysis ◽  
High Performance ◽  
State Of The Art ◽  
Thermoplastic Composite ◽  
Light Weight

scholarly journals High-Performance Emerging Solid-State Memory Technologies

MRS Bulletin ◽  
2004 ◽  
Vol 29 (11) ◽  
pp. 805-813 ◽  
Author(s):  
Herb Goronkin ◽  
Yang Yang
Keyword(s):  
Solid State ◽  
Research And Development ◽  
High Performance ◽  
New Technologies ◽  
State Of The Art ◽  
Electronic Equipment ◽  
Private Industry ◽  
Development Stages ◽  
The World ◽  
And Storage

AbstractThis article introduces the November 2004 issue of MRS Bulletin on the state of the art in solid-state memory and storage technologies.The memory business drives hundreds of billions of dollars in sales of electronic equipment per year. The incentive for continuing on the historical track outlined by Moore's law is huge, and this challenge is driving considerable investment from governments around the world as well as in private industry and universities. The problem is this: recognizing that current approaches to semiconductor-based memory are limited, what new technologies can be introduced to continue or even accelerate the pace of complexity? The articles in this issue highlight several commercially available memories, as well as memory technologies that are still in the research and development stages. What will become apparent to the reader is the huge diversity of approaches to this problem.

scholarly journals An Effective Algorithm for Denoising Salt-And-Pepper Noise in Real-Time

2020 ◽  
pp. 14-27
Author(s):  
Obed Appiah ◽  
James Benjamin Hayfron-Acquah ◽  
Michael Asante
Keyword(s):  
Real Time ◽  
High Performance ◽  
State Of The Art ◽  
Median Filter ◽  
Image Data ◽  
Poor Performance ◽  
Relevant Information ◽  
Salt And Pepper Noise ◽  
Running Time ◽  
Salt And Pepper

For computer vision systems to effectively perform diagnoses, identification, tracking, monitoring and surveillance, image data must be devoid of noise. Various types of noises such as Salt-and-pepper or Impulse, Gaussian, Shot, Quantization, Anisotropic, and Periodic noises corrupts images making it difficult to extract relevant information from them. This has led to a lot of proposed algorithms to help fix the problem. Among the proposed algorithms, the median filter has been successful in handling salt-and-pepper noise and preserving edges in images. However, its moderate to high running time and poor performance when images are corrupted with high densities of noise, has led to various proposed modifications of the median filter. The challenge observed with all these modifications is the trade-off between efficient running time and quality of denoised images. This paper proposes an algorithm that delivers quality denoised images in low running time. Two state-of-the-art algorithms are combined into one and a technique called Mid-Value-Decision-Median introduced into the proposed algorithm to deliver high quality denoised images in real-time. The proposed algorithm, High-Performance Modified Decision Based Median Filter (HPMDBMF) runs about 200 times faster than the state-of-the-art Modified Decision Based Median Filter (MDBMF) and still generate equivalent output.

scholarly journals PlaneNet: an efficient local feature extraction network

2021 ◽  
Vol 7 ◽  
pp. e783
Author(s):  
Bin Lin ◽  
Houcheng Su ◽  
Danyang Li ◽  
Ao Feng ◽  
Hongxiang Li ◽  
...  
Keyword(s):  
Mobile Devices ◽  
State Of The Art ◽  
Computing Time ◽  
Semantic Segmentation ◽  
Vital Role ◽  
Light Weight ◽  
Practical Application ◽  
Classification Tasks ◽  
Segmentation Task ◽  
Local Feature Extraction

Due to memory and computing resources limitations, deploying convolutional neural networks on embedded and mobile devices is challenging. However, the redundant use of the 1 × 1 convolution in traditional light-weight networks, such as MobileNetV1, has increased the computing time. By utilizing the 1 × 1 convolution that plays a vital role in extracting local features more effectively, a new lightweight network, named PlaneNet, is introduced. PlaneNet can improve the accuracy and reduce the numbers of parameters and multiply-accumulate operations (Madds). Our model is evaluated on classification and semantic segmentation tasks. In the classification tasks, the CIFAR-10, Caltech-101, and ImageNet2012 datasets are used. In the semantic segmentation task, PlaneNet is tested on the VOC2012 datasets. The experimental results demonstrate that PlaneNet (74.48%) can obtain higher accuracy than MobileNetV3-Large (73.99%) and GhostNet (72.87%) and achieves state-of-the-art performance with fewer network parameters in both tasks. In addition, compared with the existing models, it has reached the practical application level on mobile devices. The code of PlaneNet on GitHub: https://github.com/LinB203/planenet.

scholarly journals Emerging MXene–Polymer Hybrid Nanocomposites for High-Performance Ammonia Sensing and Monitoring

Nanomaterials ◽  
2021 ◽  
Vol 11 (10) ◽  
pp. 2496
Author(s):  
Vishal Chaudhary ◽  
Akash Gautam ◽  
Yogendra K. Mishra ◽  
Ajeet Kaushik
Keyword(s):  
Biomedical Applications ◽  
High Performance ◽  
State Of The Art ◽  
Hybrid Nanocomposites ◽  
Global Public Health ◽  
Ammonia Sensing ◽  
Polymer Hybrid ◽  
And Storage ◽  
Sensing Characteristics ◽  
Tunable Morphology

Ammonia (NH3) is a vital compound in diversified fields, including agriculture, automotive, chemical, food processing, hydrogen production and storage, and biomedical applications. Its extensive industrial use and emission have emerged hazardous to the ecosystem and have raised global public health concerns for monitoring NH3 emissions and implementing proper safety strategies. These facts created emergent demand for translational and sustainable approaches to design efficient, affordable, and high-performance compact NH3 sensors. Commercially available NH3 sensors possess three major bottlenecks: poor selectivity, low concentration detection, and room-temperature operation. State-of-the-art NH3 sensors are scaling up using advanced nano-systems possessing rapid, selective, efficient, and enhanced detection to overcome these challenges. MXene–polymer nanocomposites (MXP-NCs) are emerging as advanced nanomaterials of choice for NH3 sensing owing to their affordability, excellent conductivity, mechanical flexibility, scalable production, rich surface functionalities, and tunable morphology. The MXP-NCs have demonstrated high performance to develop next-generation intelligent NH3 sensors in agricultural, industrial, and biomedical applications. However, their excellent NH3-sensing features are not articulated in the form of a review. This comprehensive review summarizes state-of-the-art MXP-NCs fabrication techniques, optimization of desired properties, enhanced sensing characteristics, and applications to detect airborne NH3. Furthermore, an overview of challenges, possible solutions, and prospects associated with MXP-NCs is discussed.

scholarly journals Efficient and Effective Incomplete Multi-View Clustering

2019 ◽  
Vol 33 ◽  
pp. 4392-4399 ◽  
Author(s):  
Xinwang Liu ◽  
Xinzhong Zhu ◽  
Miaomiao Li ◽  
Chang Tang ◽  
En Zhu ◽  
...  
Keyword(s):  
Computational Complexity ◽  
Optimization Problem ◽  
State Of The Art ◽  
Joint Optimization ◽  
Kernel Matrix ◽  
Consensus Clustering ◽  
Running Time ◽  
Multiple Kernel ◽  
Base Matrix ◽  
And Storage

Incomplete multi-view clustering (IMVC) optimally fuses multiple pre-specified incomplete views to improve clustering performance. Among various excellent solutions, the recently proposed multiple kernel k-means with incomplete kernels (MKKM-IK) forms a benchmark, which redefines IMVC as a joint optimization problem where the clustering and kernel matrix imputation tasks are alternately performed until convergence. Though demonstrating promising performance in various applications, we observe that the manner of kernel matrix imputation in MKKM-IK would incur intensive computational and storage complexities, overcomplicated optimization and limitedly improved clustering performance. In this paper, we propose an Efficient and Effective Incomplete Multi-view Clustering (EE-IMVC) algorithm to address these issues. Instead of completing the incomplete kernel matrices, EE-IMVC proposes to impute each incomplete base matrix generated by incomplete views with a learned consensus clustering matrix. We carefully develop a three-step iterative algorithm to solve the resultant optimization problem with linear computational complexity and theoretically prove its convergence. Further, we conduct comprehensive experiments to study the proposed EE-IMVC in terms of clustering accuracy, running time, evolution of the learned consensus clustering matrix and the convergence. As indicated, our algorithm significantly and consistently outperforms some state-of-the-art algorithms with much less running time and memory.

Sign in / Sign up

Export Citation Format

Share Document