scholarly journals Foreground Detection with Deeply Learned Multi-Scale Spatial-Temporal Features

Sensors ◽  
2018 ◽  
Vol 18 (12) ◽  
pp. 4269 ◽  
Author(s):  
Yao Wang ◽  
Zujun Yu ◽  
Liqiang Zhu
Keyword(s):  
Neural Network ◽  
Moving Objects ◽  
Fundamental Problem ◽  
Image Features ◽  
Foreground Detection ◽  
Training Procedure ◽  
Spatial And Temporal Scales ◽  
Multi Scale ◽  
Temporal Features ◽  
Background Models

Foreground detection, which extracts moving objects from videos, is an important and fundamental problem of video analysis. Classic methods often build background models based on some hand-craft features. Recent deep neural network (DNN) based methods can learn more effective image features by training, but most of them do not use temporal feature or use simple hand-craft temporal features. In this paper, we propose a new dual multi-scale 3D fully-convolutional neural network for foreground detection problems. It uses an encoder–decoder structure to establish a mapping from image sequences to pixel-wise classification results. We also propose a two-stage training procedure, which trains the encoder and decoder separately to improve the training results. With multi-scale architecture, the network can learning deep and hierarchical multi-scale features in both spatial and temporal domains, which is proved to have good invariance for both spatial and temporal scales. We used the CDnet dataset, which is currently the largest foreground detection dataset, to evaluate our method. The experiment results show that the proposed method achieves state-of-the-art results in most test scenes, comparing to current DNN based methods.

scholarly journals Fault Diagnosis of Main Pump in Converter Station Based on Deep Neural Network

Symmetry ◽  
2021 ◽  
Vol 13 (7) ◽  
pp. 1284
Author(s):  
Qingsheng Zhao ◽  
Gong Cheng ◽  
Xiaoqing Han ◽  
Dingkang Liang ◽  
Xuping Wang
Keyword(s):  
Neural Network ◽  
Fault Diagnosis ◽  
Short Term Memory ◽  
Cooling System ◽  
Imbalanced Data ◽  
Maintenance Cost ◽  
Stable Operation ◽  
Vibration Signals ◽  
Multi Scale ◽  
Temporal Features

As the core component of the valve cooling system in a converter station, the main pump plays a major role in ensuring the stable operation of the valve. Thus, accurate and efficient fault diagnosis of the main pump according to vibration signals is of positive significance for the detection of failure equipment and reducing the maintenance cost. This paper proposed a new neural network based on the vibration signals of the main pump to classify four faults and one normal state of the main pump, which consisted of a convolutional neural network (CNN) and long short-term memory (LSTM). Multi-scale features were extracted by two CNNs with different kernel sizes, and temporal features were extracted by LSTM. Moreover, random sampling was used in data processing for imbalanced data, which is meaningful for data symmetry. Experimental results indicated that the accuracy of the network was 0.987 obtained from the test set, and the average values of F1-score, recall, and precision were 0.987, 0.987, and 0.988, respectively. It was found that the proposed network performed well in a multi-label fault diagnosis of the main pump and was superior to other methods.

scholarly journals Super-resolution reconstruction of seismic section image via multi-scale convolution neural network

2021 ◽  
Vol 303 ◽  
pp. 01058
Author(s):  
Meng-Di Deng ◽  
Rui-Sheng Jia ◽  
Hong-Mei Sun ◽  
Xing-Li Zhang
Keyword(s):  
Neural Network ◽  
High Resolution ◽  
Convolutional Neural Network ◽  
Super Resolution ◽  
Image Features ◽  
Image Feature ◽  
Reconstruction Method ◽  
Low Resolution ◽  
Seismic Section ◽  
Multi Scale

The resolution of seismic section images can directly affect the subsequent interpretation of seismic data. In order to improve the spatial resolution of low-resolution seismic section images, a super-resolution reconstruction method based on multi-scale convolution is proposed. This method designs a multi-scale convolutional neural network to learn high-low resolution image feature pairs, and realizes mapping learning from low-resolution seismic section images to high-resolution seismic section images. This multi-scale convolutional neural network model consists of four convolutional layers and a sub-pixel convolutional layer. Convolution operations are used to learn abundant seismic section image features, and sub-pixel convolution layer is used to reconstruct high-resolution seismic section image. The experimental results show that the proposed method is superior to the comparison method in peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). In the total training time and reconstruction time, our method is about 22% less than the FSRCNN method and about 18% less than the ESPCN method.

scholarly journals DYNAMIC HANDWRITTEN SIGNATURE IDENTIFICATION USING SPIKING NEURAL NETWORK

2021 ◽  
Vol 11 (3) ◽  
pp. 34-39
Author(s):  
Vladislav Kutsman ◽  
Oleh Kolesnytskyj
Keyword(s):  
Neural Network ◽  
Slope Angle ◽  
Experimental Studies ◽  
Spiking Neural Network ◽  
Training Procedure ◽  
Spatial And Temporal Scales ◽  
Highly Correlated ◽  
The One ◽  
Dynamic Signature ◽  
Handwritten Signature

The article proposes a method for dynamic signature identification based on a spiking neural network. Three dynamic signature parameters l(t), xy(t), p(t) are used, which are invariant to the signature slope angle, and after their normalization, also to the signature spatial and temporal scales. These dynamic parameters are fed to the spiking neural network for recognition simultaneously in the form of time series without preliminary transformation into a vector of static features, which, on the one hand, simplifies the method due to the absence of complex computational transformation procedures, and on the other hand, prevents the loss of useful information, and therefore increases the accuracy and reliability of signature identification and recognition (especially when recognizing forged signatures that are highly correlated with the genuine). The spiking neural network used has a simple training procedure, and not all neurons of the network are trained, but only the output ones. If it is necessary to add new signatures, it is not necessary to retrain the entire network as a whole, but it is enough to add several output neurons and learn only their connections. In the results of experimental studies of the software implementation of the proposed system, it’s EER = 3.9% was found when identifying skilled forgeries and EER = 0.17% when identifying random forgeries.

Automatic Recognition of Flock Behavior of Chickens with Convolutional Neural Network and Kinect Sensor

2018 ◽  
Vol 32 (07) ◽  
pp. 1850023 ◽  
Author(s):  
Haitao Pu ◽  
Jian Lian ◽  
Mingqu Fan
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Learning Strategy ◽  
State Of The Art ◽  
Domestic Animal ◽  
Image Features ◽  
Kinect Sensor ◽  
Multi Scale ◽  
Art Methods ◽  
Effectiveness And Efficiency

In this paper, we propose an automatic convolutional neural network (CNN)-based method to recognize the chicken behavior within a poultry farm using a Kinect sensor. It resolves the hardships in flock behavior image classification by leveraging a data-driven mechanism and exploiting non-manually extracted multi-scale image features which combine both the local and global characteristics of the image. To our best knowledge, this is probably the first attempt of deep learning strategy in the field of domestic animal behavior recognition. To testify the performance of our proposed method, we conducted experiments between state-of-the-art methods and our method. Experimental results witness that our proposed approach outperforms the state-of-the-art methods both in effectiveness and efficiency. Our proposed CNN architecture for recognizing flock behavior of chickens produces an extremely impressive accuracy of 99.17%.

scholarly journals MARnet: multi-scale adaptive residual neural network for chest X-ray images recognition of lung diseases

2021 ◽  
Vol 19 (1) ◽  
pp. 331-350
Author(s):  
Boyang Wang ◽  
◽  
Wenyu Zhang
Keyword(s):  
Neural Network ◽  
Lung Diseases ◽  
Rapid Development ◽  
Image Features ◽  
Learning Technology ◽  
X Ray ◽  
Multi Scale ◽  
Clinical Diagnostic ◽  
Chest X Ray ◽  
Images Recognition

<abstract> <p>Chest X-ray image is an important clinical diagnostic reference to lung diseases that is a serious threat to human health. At present, with the rapid development of computer vision and deep learning technology, many scholars have carried out the fruitful research on how to build a valid model for chest X-ray images recognition of lung diseases. While some efforts are still expected to improve the performance of the recognition model and enhance the interpretability of the recognition results. In this paper, we construct a multi-scale adaptive residual neural network (MARnet) to identify chest X-ray images of lung diseases. To make the model better extract image features, we cross-transfer the information extracted by residual block and the information extracted by adaptive structure to different layer, avoiding the reduction effect of residual structure on adaptive function. We compare MARnet with some classical neural networks, and the results show that MARnet achieves accuracy (ACC) of 83.3% and the area under ROC curve (AUC) of 0.97 in the identification of 4 kinds of typical lung X-ray images including nodules, atelectasis, normal and infection, which are higher than those of other methods. Moreover, to avoid the randomness of the train-test-split method, 5-fold cross-validation method is used to verify the generalization ability of the MARnet model and the results are satisfactory. Finally, the technique called Gradient-weighted Class Activation Mapping (Grad-CAM), is adopted to display significantly the discriminative regions of the images in the form of the heat map, which provides an explainable and more direct clinical diagnostic reference to lung diseases.</p> </abstract>

Digitalization system of ancient architecture decoration art based on neural network and image features

2020 ◽  
pp. 1-12
Author(s):  
Wu Xin ◽  
Qiu Daping
Keyword(s):  
Neural Network ◽  
Construction Industry ◽  
Three Dimensional ◽  
Performance Testing ◽  
Image Features ◽  
Three Dimensional Model ◽  
Performance Effect ◽  
Data Process ◽  
And Performance ◽  
Construction Mode

The inheritance and innovation of ancient architecture decoration art is an important way for the development of the construction industry. The data process of traditional ancient architecture decoration art is relatively backward, which leads to the obvious distortion of the digitalization of ancient architecture decoration art. In order to improve the digital effect of ancient architecture decoration art, based on neural network, this paper combines the image features to construct a neural network-based ancient architecture decoration art data system model, and graphically expresses the static construction mode and dynamic construction process of the architecture group. Based on this, three-dimensional model reconstruction and scene simulation experiments of architecture groups are realized. In order to verify the performance effect of the system proposed in this paper, it is verified through simulation and performance testing, and data visualization is performed through statistical methods. The result of the study shows that the digitalization effect of the ancient architecture decoration art proposed in this paper is good.

scholarly journals Multi-Scale Aggregation Graph Neural Networks Based on Feature Similarity for Semi-Supervised Learning

Entropy ◽  
2021 ◽  
Vol 23 (4) ◽  
pp. 403
Author(s):  
Xun Zhang ◽  
Lanyan Yang ◽  
Bin Zhang ◽  
Ying Liu ◽  
Dong Jiang ◽  
...  
Keyword(s):  
Neural Network ◽  
Supervised Learning ◽  
Structured Data ◽  
Similarity Matrix ◽  
Scale Feature ◽  
Multi Scale ◽  
Adjacent Node ◽  
Feature Similarity ◽  
The Mean ◽  
Graph Neural Networks

The problem of extracting meaningful data through graph analysis spans a range of different fields, such as social networks, knowledge graphs, citation networks, the World Wide Web, and so on. As increasingly structured data become available, the importance of being able to effectively mine and learn from such data continues to grow. In this paper, we propose the multi-scale aggregation graph neural network based on feature similarity (MAGN), a novel graph neural network defined in the vertex domain. Our model provides a simple and general semi-supervised learning method for graph-structured data, in which only a very small part of the data is labeled as the training set. We first construct a similarity matrix by calculating the similarity of original features between all adjacent node pairs, and then generate a set of feature extractors utilizing the similarity matrix to perform multi-scale feature propagation on graphs. The output of multi-scale feature propagation is finally aggregated by using the mean-pooling operation. Our method aims to improve the model representation ability via multi-scale neighborhood aggregation based on feature similarity. Extensive experimental evaluation on various open benchmarks shows the competitive performance of our method compared to a variety of popular architectures.

Sign in / Sign up

Export Citation Format

Share Document