Global and Local Knowledge-Aware Attention Network for Action Recognition

Zheng, Zhenxing; An, Gaoyun; Wu, Dapeng; Ruan, Qiuqi

doi:10.1109/tnnls.2020.2978613

Cited by 44 publications

(18 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, we have combined PSE block and SE block in the model for better performance. In the future, we will study how to use PSE blocks for more complex tasks [46][47][48], and use PSigmoid in other networks [49][50][51][52].…”

Section: Resultsmentioning

confidence: 99%

PSigmoid: Improving squeeze-and-excitation block with parametric sigmoid

et al. 2021

View full text Add to dashboard Cite

Squeeze-and-Excitation (SE) Networks won the last ImageNet Large-Scale Visual Recognition Challenge (ILSVRC) classification competition and is very popular in today's vision community. The SE block is the core of Squeeze-and-Excitation Network (SENet), which adaptively recalibrates channel-wise features and suppresses less useful ones. Since SE blocks can be directly used in existing models and effectively improve performance, SE blocks are widely used in a variety of tasks. In this paper, we propose a novel Parametric Sigmoid (PSigmoid) to enhance the SE block. We named the new module PSigmoid SE (PSE) block. The PSE block can not only suppress features in a channel-wise manner, but also enhance features. We evaluate the performance of our method on four common datasets including CIFAR-10, CIFAR-100, SVHN and Tiny ImageNet. Experimental results show the effectiveness of our method. In addition, we compare the differences between the PSE block and the SE block through a detailed analysis of the configuration. Finally, we use a combination of PSE block and SE block to obtain better performance.

show abstract

Section: Resultsmentioning

confidence: 99%

PSigmoid: Improving squeeze-and-excitation block with parametric sigmoid

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The authors of [ 54 ] proposed a three streams attention network for activity detection. These were statistic-based, learning-based and global-pooling attention streams.…”

Section: Related Workmentioning

confidence: 99%

A Deep Learning Framework for Recognizing Both Static and Dynamic Gestures

Mazhar

Ramdani

Cherubini

2021

Sensors

View full text Add to dashboard Cite

Intuitive user interfaces are indispensable to interact with the human centric smart environments. In this paper, we propose a unified framework that recognizes both static and dynamic gestures, using simple RGB vision (without depth sensing). This feature makes it suitable for inexpensive human-robot interaction in social or industrial settings. We employ a pose-driven spatial attention strategy, which guides our proposed Static and Dynamic gestures Network—StaDNet. From the image of the human upper body, we estimate his/her depth, along with the region-of-interest around his/her hands. The Convolutional Neural Network (CNN) in StaDNet is fine-tuned on a background-substituted hand gestures dataset. It is utilized to detect 10 static gestures for each hand as well as to obtain the hand image-embeddings. These are subsequently fused with the augmented pose vector and then passed to the stacked Long Short-Term Memory blocks. Thus, human-centred frame-wise information from the augmented pose vector and from the left/right hands image-embeddings are aggregated in time to predict the dynamic gestures of the performing person. In a number of experiments, we show that the proposed approach surpasses the state-of-the-art results on the large-scale Chalearn 2016 dataset. Moreover, we transfer the knowledge learned through the proposed methodology to the Praxis gestures dataset, and the obtained results also outscore the state-of-the-art on this dataset.

show abstract

“…Deep learning can achieve complex function approximation by learning a group of kernel parameters of a nonlinear network, which shows a strong ability to learn the essential characteristics from a small sample set. CNN is a multilayer feedforward neural network with a convolution structure and becomes one of the most concerned research hotspots (Wang et al, 2019;Zheng et al, 2020). Unlike the traditional fully connected feedforward neural network, CNN has the characteristics of local connection and parameter sharing, which reduces the complexity of the network and improves computational efficiency.…”

Section: Convolution Neural Networkmentioning

confidence: 99%

“…As a new research direction in machine learning, deep learning has attracted growing attentions (Zheng et al, 2020). CNN is a kind of multilayer feedforward neural network with convolution operations.…”

Section: Introductionmentioning

confidence: 99%

A Novel Shale Gas Production Prediction Model Based on Machine Learning and Its Application in Optimization of Multistage Fractured Horizontal Wells

Wang

Qiao

et al. 2021

Front. Earth Sci.

View full text Add to dashboard Cite

Shale gas production prediction and horizontal well parameter optimization are significant for shale gas development. However, conventional reservoir numerical simulation requires extensive resources in terms of labor, time, and computations, and so the optimization problem still remains a challenge. Therefore, we propose, for the first time, a new gas production prediction methodology based on Gaussian Process Regression (GPR) and Convolution Neural Network (CNN) to complement the numerical simulation model and achieve rapid optimization. Specifically, through sensitivity analysis, porosity, permeability, fracture half-length, and horizontal well length were selected as influencing factors. Second, the n-factorial experimental design was applied to design the initial experiment and the dataset was constructed by combining the simulation results with the case parameters. Subsequently, the gas production model was built by GPR, CNN, and SVM based on the dataset. Finally, the optimal model was combined with the optimization algorithm to maximize the Net Present Value (NPV) and obtain the optimal fracture half-length and horizontal well length. Experimental results demonstrated the GPR model had prominent modeling capabilities compared with CNN and Support Vector Machine (SVM) and achieved the satisfactory prediction performance. The fracture half-length and well length optimized by the GPR model and reservoir numerical simulation model converged to almost the same values. Compared with the field reference case, the optimized NPV increased by US$ 7.43 million. Additionally, the time required to optimize the GPR model was 1/720 of that of numerical simulation. This work enriches the knowledge of shale gas development technology and lays the foundation for realizing the scale-benefit development for shale gas, so as to realize the integration of geological engineering.

show abstract

Global and Local Knowledge-Aware Attention Network for Action Recognition

Cited by 44 publications

References 55 publications

PSigmoid: Improving squeeze-and-excitation block with parametric sigmoid

PSigmoid: Improving squeeze-and-excitation block with parametric sigmoid

A Deep Learning Framework for Recognizing Both Static and Dynamic Gestures

A Novel Shale Gas Production Prediction Model Based on Machine Learning and Its Application in Optimization of Multistage Fractured Horizontal Wells

Contact Info

Product

Resources

About