Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824686
Wanbo Luo
Many manufacturers will strictly control the quality of products, especially the surface quality of products. Under the same conditions, the better the surface quality of the product, the more competitive it is. Many aluminum profile benchmarking companies have pain points with flaws on the surface of their products. Due to the work mistakes of the workers in the production workshop, unqualified aluminum materials need to be eliminated in the product production control, and the traditional method is to rely on the assembly line workers to check one by one. As the company’s production automation continues to increase, the shortcomings of manual inspection methods have become increasingly prominent. Aiming at the common types of surface defects in the company’s aluminum profile production process, this paper introduces the deep learning method into the identification of aluminum profile surface defects and uses convolutional neural network to identify the surface defects of aluminum profiles. The advantages and disadvantages of different aluminum profile surface defect recognition models such as AlexNet, VGG19 and Inception V4 are analyzed. Finally, according to the recognition effect of the aluminum profile data set, the recognition model of aluminum profile surface defects based on Inception V4 is selected as the optimal model.
{"title":"Recognition of surface defects of aluminum profiles based on convolutional neural network","authors":"Wanbo Luo","doi":"10.1109/cvidliccea56201.2022.9824686","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824686","url":null,"abstract":"Many manufacturers will strictly control the quality of products, especially the surface quality of products. Under the same conditions, the better the surface quality of the product, the more competitive it is. Many aluminum profile benchmarking companies have pain points with flaws on the surface of their products. Due to the work mistakes of the workers in the production workshop, unqualified aluminum materials need to be eliminated in the product production control, and the traditional method is to rely on the assembly line workers to check one by one. As the company’s production automation continues to increase, the shortcomings of manual inspection methods have become increasingly prominent. Aiming at the common types of surface defects in the company’s aluminum profile production process, this paper introduces the deep learning method into the identification of aluminum profile surface defects and uses convolutional neural network to identify the surface defects of aluminum profiles. The advantages and disadvantages of different aluminum profile surface defect recognition models such as AlexNet, VGG19 and Inception V4 are analyzed. Finally, according to the recognition effect of the aluminum profile data set, the recognition model of aluminum profile surface defects based on Inception V4 is selected as the optimal model.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"56 8 1","pages":"5-8"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89687454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For the problem that ceramic ring defects are small and difficult to detect with many types; and the defect feature information is weak and difficult to extract, this paper proposes an improved YOLOv5-based target detection method to achieve the detection of ceramic ring defects. By adding an attention mechanism to the Backbone part of YOLOv5, the attention of the network model to different types of defects can be improved, the interference of irrelevant background can be reduced, and the network can extract the channel features and spatial features of the defects more effectively, which can effectively enhance the detection capability of the model. The experimental results show that the ceramic ring defect detection method proposed in this paper can accurately detect defects with an mAP value of 89.9%, which is 1.1% better compared with the original YOLOv5 algorithm. It provides an effective detection method for defect detection of ceramic ring parts.
{"title":"Ceramic ring defect detection based on improved YOLOv5","authors":"Shengqi Guan, Xu Wang, Jingguo Wang, Zijiang Yu, Xizhi Wang, Chao Zhang, Tong Liu, Dongdong Liu, Junqiang Wang, Libo Zhang","doi":"10.1109/cvidliccea56201.2022.9824099","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824099","url":null,"abstract":"For the problem that ceramic ring defects are small and difficult to detect with many types; and the defect feature information is weak and difficult to extract, this paper proposes an improved YOLOv5-based target detection method to achieve the detection of ceramic ring defects. By adding an attention mechanism to the Backbone part of YOLOv5, the attention of the network model to different types of defects can be improved, the interference of irrelevant background can be reduced, and the network can extract the channel features and spatial features of the defects more effectively, which can effectively enhance the detection capability of the model. The experimental results show that the ceramic ring defect detection method proposed in this paper can accurately detect defects with an mAP value of 89.9%, which is 1.1% better compared with the original YOLOv5 algorithm. It provides an effective detection method for defect detection of ceramic ring parts.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"1 1","pages":"115-118"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89892765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824758
Xiao Wang, Bo Zhao, Shengxian Cao, Siyuan Fan
In this paper, a novel multi-sensor information fusion (MSIF) monitoring system of photovoltaic (PV) power station is proposed, which can solve the difficulty in determining the dust accumulation degree of PV power station operation and maintenance personnel to the panels. According to the real-time monitoring data, a relationship model can be established to reflect the effect of dust accumulation on PV panels operating state. Meanwhile, a dust detection and classification method based on convolutional neural network (CNN) is also given to analyze the visible-light images and operation data. Because of identifying rapidly the images of the dust-covered PV panels, the classification result can be used as a guideline for cleaning the dust accumulation of PV panels. Finally, the experimental results show the effectiveness of the proposed monitoring system.
{"title":"A multi-sensor information fusion monitoring system for photovoltaic power generation","authors":"Xiao Wang, Bo Zhao, Shengxian Cao, Siyuan Fan","doi":"10.1109/cvidliccea56201.2022.9824758","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824758","url":null,"abstract":"In this paper, a novel multi-sensor information fusion (MSIF) monitoring system of photovoltaic (PV) power station is proposed, which can solve the difficulty in determining the dust accumulation degree of PV power station operation and maintenance personnel to the panels. According to the real-time monitoring data, a relationship model can be established to reflect the effect of dust accumulation on PV panels operating state. Meanwhile, a dust detection and classification method based on convolutional neural network (CNN) is also given to analyze the visible-light images and operation data. Because of identifying rapidly the images of the dust-covered PV panels, the classification result can be used as a guideline for cleaning the dust accumulation of PV panels. Finally, the experimental results show the effectiveness of the proposed monitoring system.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"45 1","pages":"955-959"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89412823","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824921
Xinjun Zhu, Zhizhi Zhang, Linpeng Hou, Limei Song, Hongyi Wang
Light field structured light 3D measurement has gained popularity by merging the advantages of light field and structured light methods. Generating light field structured light dataset is necessary for studying light field 3D reconstruction algorithms, but it is time-consuming and expensive in a real sense, especially for ground truth data. This paper proposes a method to generate light field structured light projection data with Blender simulation. The proposed method allows for the modification of camera and projector settings and parameters, as well as rotating objects. The dataset generated by this method contains 107730 light field structured light images. The label data (ground truth data) including depth and disparity by the 9×9 light field camera array are provided for the performance evaluation of 3D reconstruction algorithms. To the best of our knowledge, it is the first public dataset in the light field structured light projection environment. Diverse 3D reconstruction methods, including deep learning methods, are used to evaluate the proposed data generation method and dataset. The dataset is available at https://github.com/sabaizzz/Light-field-structured-light-dataset.
光场结构光三维测量融合了光场和结构光两种方法的优点，得到了广泛的应用。生成光场结构光数据集是研究光场三维重建算法的必要条件，但实际意义上的光场结构光数据集既耗时又昂贵，特别是对于地真数据集更是如此。本文提出了一种利用Blender模拟生成光场结构光投影数据的方法。所提出的方法允许修改相机和投影仪的设置和参数，以及旋转对象。该方法生成的数据集包含107730张光场结构光图像。通过9×9光场相机阵列提供包含深度和视差的标签数据(ground truth data)，用于三维重建算法的性能评估。据我们所知，这是光场结构光投影环境中的第一个公开数据集。不同的三维重建方法，包括深度学习方法，被用来评估提出的数据生成方法和数据集。该数据集可在https://github.com/sabaizzz/Light-field-structured-light-dataset上获得。
{"title":"Light field structured light projection data generation with Blender","authors":"Xinjun Zhu, Zhizhi Zhang, Linpeng Hou, Limei Song, Hongyi Wang","doi":"10.1109/cvidliccea56201.2022.9824921","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824921","url":null,"abstract":"Light field structured light 3D measurement has gained popularity by merging the advantages of light field and structured light methods. Generating light field structured light dataset is necessary for studying light field 3D reconstruction algorithms, but it is time-consuming and expensive in a real sense, especially for ground truth data. This paper proposes a method to generate light field structured light projection data with Blender simulation. The proposed method allows for the modification of camera and projector settings and parameters, as well as rotating objects. The dataset generated by this method contains 107730 light field structured light images. The label data (ground truth data) including depth and disparity by the 9×9 light field camera array are provided for the performance evaluation of 3D reconstruction algorithms. To the best of our knowledge, it is the first public dataset in the light field structured light projection environment. Diverse 3D reconstruction methods, including deep learning methods, are used to evaluate the proposed data generation method and dataset. The dataset is available at https://github.com/sabaizzz/Light-field-structured-light-dataset.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"14 1","pages":"1249-1253"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78143540","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824849
Jingyao Nie, Shuqin Geng, Xiaohong Peng, Wenhua Cao, Pengkun Li, Xuefeng Li
Dynamic verification is used extensively in making sure the logical correctness of design. Coverage is often used to measure the progress of the current verification. After the coverage criteria has been met, there may still be potential bugs that have not been detected. This paper proposes a method to help engineers analyze where the potential bug is most likely to occur. We construct a cover framework and propose an algorithm to calculate the probability. Experimental results based on Monte Carlo simulation are agreed with our algorithm.
{"title":"A Location Analysis for Dynamic Verification","authors":"Jingyao Nie, Shuqin Geng, Xiaohong Peng, Wenhua Cao, Pengkun Li, Xuefeng Li","doi":"10.1109/cvidliccea56201.2022.9824849","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824849","url":null,"abstract":"Dynamic verification is used extensively in making sure the logical correctness of design. Coverage is often used to measure the progress of the current verification. After the coverage criteria has been met, there may still be potential bugs that have not been detected. This paper proposes a method to help engineers analyze where the potential bug is most likely to occur. We construct a cover framework and propose an algorithm to calculate the probability. Experimental results based on Monte Carlo simulation are agreed with our algorithm.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"116 1","pages":"524-527"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73875107","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9825336
Zhisong Bai, Chao Zhang, Cheng Han, Linke Zhang
ASPP (Atrous Spatial Pooling Pyramid) has the advantage that it can expand the receptive field and extract multiscale features without changing the image resolution. We introduce it into the depth estimation task to improve the problems of inaccurate depth estimation, blurred edges, and loss of depth information details on the current unsupervised depth estimation methods for indoor monocular images. However, the ASPP module does not consider the relationship between different pixel features, resulting in inaccurate extraction of scene features in the depth estimation task. Therefore, we propose a Tree-shaped ASPP structure for this drawback, combined with the SC-SfMLearner network using the NYUv2 dataset, adding the spatial semantic information pool formed by the ASPP tree structure between the encoder and decoder structures of the depth estimation network, which can not only expand the receptive field without losing resolution but also capture and fuse multi-scale context information, so that different pixels establish connections between features. The results show that, compared with the original method, the improved method has stronger network feature extraction ability, clearer contours of each target in the scene, more distinct layers, and more accurate depth estimation results.
{"title":"Indoor Monocular Image Depth Estimation Based on Semantic Information of Tree-shaped ASPP Structure","authors":"Zhisong Bai, Chao Zhang, Cheng Han, Linke Zhang","doi":"10.1109/cvidliccea56201.2022.9825336","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9825336","url":null,"abstract":"ASPP (Atrous Spatial Pooling Pyramid) has the advantage that it can expand the receptive field and extract multiscale features without changing the image resolution. We introduce it into the depth estimation task to improve the problems of inaccurate depth estimation, blurred edges, and loss of depth information details on the current unsupervised depth estimation methods for indoor monocular images. However, the ASPP module does not consider the relationship between different pixel features, resulting in inaccurate extraction of scene features in the depth estimation task. Therefore, we propose a Tree-shaped ASPP structure for this drawback, combined with the SC-SfMLearner network using the NYUv2 dataset, adding the spatial semantic information pool formed by the ASPP tree structure between the encoder and decoder structures of the depth estimation network, which can not only expand the receptive field without losing resolution but also capture and fuse multi-scale context information, so that different pixels establish connections between features. The results show that, compared with the original method, the improved method has stronger network feature extraction ability, clearer contours of each target in the scene, more distinct layers, and more accurate depth estimation results.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"56 1","pages":"330-333"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84558711","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824420
J. Sun, Neng He, Jiawen Zhang, Huameng Gao, Fenling Qi, Hongjiang Yang
Studying the distribution law of equipment failure is the foundation of reliability engineering research. In order to study the problems of the complex structure and failure mechanism of the wireless communication hardware system, and the difficulty in describing the overall reliability, this paper starts from the research on the failure distribution of the underlying hardware equipment. By comparing and analyzing the correlation coefficient and error value of the exponential distribution of hardware equipment life, log-normal distribution, and two-parameter Weibull distribution, it is proposed that the wireless communication hardware equipment conforms to the two-parameter Weibull failure distribution law. In this paper, combined with the operation process and failure criteria of the hardware system, a system reliability evaluation model combining series and weight connection is constructed to conduct a comprehensive evaluation of the hardware system reliability, and to test and analyze it in combination with actual cases.
{"title":"Research on reliability analysis method of wireless communication hardware system","authors":"J. Sun, Neng He, Jiawen Zhang, Huameng Gao, Fenling Qi, Hongjiang Yang","doi":"10.1109/cvidliccea56201.2022.9824420","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824420","url":null,"abstract":"Studying the distribution law of equipment failure is the foundation of reliability engineering research. In order to study the problems of the complex structure and failure mechanism of the wireless communication hardware system, and the difficulty in describing the overall reliability, this paper starts from the research on the failure distribution of the underlying hardware equipment. By comparing and analyzing the correlation coefficient and error value of the exponential distribution of hardware equipment life, log-normal distribution, and two-parameter Weibull distribution, it is proposed that the wireless communication hardware equipment conforms to the two-parameter Weibull failure distribution law. In this paper, combined with the operation process and failure criteria of the hardware system, a system reliability evaluation model combining series and weight connection is constructed to conduct a comprehensive evaluation of the hardware system reliability, and to test and analyze it in combination with actual cases.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"1 1","pages":"479-483"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84559462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824858
Zi-Hua Li
The COVID-19 epidemic is still very serious, because the United States and other countries have relaxed prevention and control, and the vaccine is ineffective against the mutant virus, resulting in a large number of new cases. The existing epidemic detection methods are still insufficient, and some detection methods are relatively expensive and complicated, resulting in the supply not keeping up with the demand for detection. The purpose of this study is to use relatively convenient, fast and low-cost computer vision technology for epidemic detection. We tried the VGG, ResNet and DenseNet models on an open Kaggle dataset, and found that DenseNet achieved the best results, achieving 95% accuracy, and there is hope for further applications in the future.
{"title":"COVID-19 Classification with CT Scan and Advanced Deep Learning Technologies","authors":"Zi-Hua Li","doi":"10.1109/cvidliccea56201.2022.9824858","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824858","url":null,"abstract":"The COVID-19 epidemic is still very serious, because the United States and other countries have relaxed prevention and control, and the vaccine is ineffective against the mutant virus, resulting in a large number of new cases. The existing epidemic detection methods are still insufficient, and some detection methods are relatively expensive and complicated, resulting in the supply not keeping up with the demand for detection. The purpose of this study is to use relatively convenient, fast and low-cost computer vision technology for epidemic detection. We tried the VGG, ResNet and DenseNet models on an open Kaggle dataset, and found that DenseNet achieved the best results, achieving 95% accuracy, and there is hope for further applications in the future.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"20 1","pages":"458-462"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84799612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9824242
Mingji Li, Ning Cao, Hao Lu, Fan Gao
In view of the fact that the operation data of metering equipment in the power system has strong nonlinearity and immediacy, traditional methods cannot effectively model, predict and analyze. This paper proposes an improved local weighted partial least squares algorithm (K-MLWPLS) based on K-means clustering. The method first uses the K-means clustering algorithm to divide the training set into k sub-training sets; then uses the locally weighted partial least squares algorithm combined with the Two-scale similarity measure to model the sub-training sets, and uses grid search and Cross-validation adjusts the model parameters to obtain the optimized k sub-models; then for the test set samples, the sub-models weighted based on the centroid neighbour’s radius are integrated to calculate the final predicted value corresponding to the test set samples. The algorithm is applied to the prediction analysis of the operation data of the metering equipment. The experimental results show that the K-MLWPLS algorithm significantly improves the prediction accuracy of the model compared with the traditional modeling algorithm.
{"title":"A Predict Method of Measuring Equipment Operation Performance Based on Improved Local Weighted Partial Least Squares","authors":"Mingji Li, Ning Cao, Hao Lu, Fan Gao","doi":"10.1109/cvidliccea56201.2022.9824242","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9824242","url":null,"abstract":"In view of the fact that the operation data of metering equipment in the power system has strong nonlinearity and immediacy, traditional methods cannot effectively model, predict and analyze. This paper proposes an improved local weighted partial least squares algorithm (K-MLWPLS) based on K-means clustering. The method first uses the K-means clustering algorithm to divide the training set into k sub-training sets; then uses the locally weighted partial least squares algorithm combined with the Two-scale similarity measure to model the sub-training sets, and uses grid search and Cross-validation adjusts the model parameters to obtain the optimized k sub-models; then for the test set samples, the sub-models weighted based on the centroid neighbour’s radius are integrated to calculate the final predicted value corresponding to the test set samples. The algorithm is applied to the prediction analysis of the operation data of the metering equipment. The experimental results show that the K-MLWPLS algorithm significantly improves the prediction accuracy of the model compared with the traditional modeling algorithm.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"12 1","pages":"41-45"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84921186","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-20DOI: 10.1109/cvidliccea56201.2022.9823997
Chaowei Zhou, J. Xiao, Aimin Xiong, Caifeng Zhang
As the global population continues to age, fall detection has become a common concern in the field of public safety. Fast and accurate detection of falling behaviors in surveillance videos and timely sending out help signals can effectively reduce the injuries caused by falls in the elderly. This paper proposes a hybrid algorithm based on an improved particle swarm optimization algorithm and a neural network for real-time fall detection in indoor environments. Human keypoints in video frames are first extracted using the alphapose model, and then the human keypoints are classified in real-time using an improved particle swarm optimization neural network model. Experimental results show that this method can effectively detect falling behaviors in indoor scenes.
{"title":"Human fall detection based on improved particle swarm optimization algorithm and neural network","authors":"Chaowei Zhou, J. Xiao, Aimin Xiong, Caifeng Zhang","doi":"10.1109/cvidliccea56201.2022.9823997","DOIUrl":"https://doi.org/10.1109/cvidliccea56201.2022.9823997","url":null,"abstract":"As the global population continues to age, fall detection has become a common concern in the field of public safety. Fast and accurate detection of falling behaviors in surveillance videos and timely sending out help signals can effectively reduce the injuries caused by falls in the elderly. This paper proposes a hybrid algorithm based on an improved particle swarm optimization algorithm and a neural network for real-time fall detection in indoor environments. Human keypoints in video frames are first extracted using the alphapose model, and then the human keypoints are classified in real-time using an improved particle swarm optimization neural network model. Experimental results show that this method can effectively detect falling behaviors in indoor scenes.","PeriodicalId":23649,"journal":{"name":"Vision","volume":"1 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2022-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79800091","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：