Pub Date : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869287
Yao Chen, Hsuan-Fu Wang
Compared to Orthogonal Frequency Division Multiplexing (OFDM), Filter Bank Multi-Carrier (FBMC) waveforms have better spectral characteristics and good interference immunity. The FBMC-OQAM technology combining Filter Bank Multi-Carrier (FBMC) and Offset Quadrature Amplitude Modulation (OQAM) has the characteristics of high spectral efficiency and no synchronization in wireless communication systems. The simulation shows that the SER performance decreases as the modulation's order increases. Furthermore, the minimum mean square error (MMSE) equalization outperforms better than the zero-forcing (ZF) equalization.1
{"title":"Perfomance Evalution for the FBMC-OQAM in the mobile fading channel","authors":"Yao Chen, Hsuan-Fu Wang","doi":"10.1109/ICCE-Taiwan55306.2022.9869287","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869287","url":null,"abstract":"Compared to Orthogonal Frequency Division Multiplexing (OFDM), Filter Bank Multi-Carrier (FBMC) waveforms have better spectral characteristics and good interference immunity. The FBMC-OQAM technology combining Filter Bank Multi-Carrier (FBMC) and Offset Quadrature Amplitude Modulation (OQAM) has the characteristics of high spectral efficiency and no synchronization in wireless communication systems. The simulation shows that the SER performance decreases as the modulation's order increases. Furthermore, the minimum mean square error (MMSE) equalization outperforms better than the zero-forcing (ZF) equalization.1","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115395134","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869088
Keita Endo, Tomotaka Kimura, Nobuhiko Itoh, T. Hiraguri
Smart agriculture has been garnering attention to improve the efficiency of works. For example, advanced technologies such as drones and Artificial Intelligence (AI) may reduce labor, increase productivity, and grow high-quality crops. The aim of our study is to photograph fields of green onions from the sky using drones, then to predict the harvest time and observe the growth situation using AI image analysis. Therefore, in this paper, we proposed basic technology for area section classification of each field by using segmentation method using deep learning to analyze the cultivation situation of each field.
{"title":"Semantic Segmentation Based Field Detection Using Drones","authors":"Keita Endo, Tomotaka Kimura, Nobuhiko Itoh, T. Hiraguri","doi":"10.1109/ICCE-Taiwan55306.2022.9869088","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869088","url":null,"abstract":"Smart agriculture has been garnering attention to improve the efficiency of works. For example, advanced technologies such as drones and Artificial Intelligence (AI) may reduce labor, increase productivity, and grow high-quality crops. The aim of our study is to photograph fields of green onions from the sky using drones, then to predict the harvest time and observe the growth situation using AI image analysis. Therefore, in this paper, we proposed basic technology for area section classification of each field by using segmentation method using deep learning to analyze the cultivation situation of each field.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117351946","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869258
Tzu-Chun Chiu, Tzung-Shi Chen, Jing-Mei Lin
This paper presents a system using Light Detection and Ranging (LiDAR) to sense the human gait, and training several deep learning models for gait recognition through the collected point cloud. Since the behavior of the human body is a continuous action, we choose deep learning architectures which deal with time-series data, Long Short-Term Memory (LSTM), Temporal Convolutional Network (TCN) and make the appropriate architecture combination to improve the accuracy of recognizing human gait.
{"title":"Human Gait Recognition using LiDAR and Deep Learning Technologies","authors":"Tzu-Chun Chiu, Tzung-Shi Chen, Jing-Mei Lin","doi":"10.1109/ICCE-Taiwan55306.2022.9869258","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869258","url":null,"abstract":"This paper presents a system using Light Detection and Ranging (LiDAR) to sense the human gait, and training several deep learning models for gait recognition through the collected point cloud. Since the behavior of the human body is a continuous action, we choose deep learning architectures which deal with time-series data, Long Short-Term Memory (LSTM), Temporal Convolutional Network (TCN) and make the appropriate architecture combination to improve the accuracy of recognizing human gait.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121102039","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}
Daily behavior irregularity detection is important for assessment of the health status for the elderly in homecare. This paper proposes a Daily Behavior Irregularity Detection (DBID) mechanism which outputs the irregularity probability of daily behaviors based on the extracted regularity features using unsupervised learning algorithm. The regular behaviors which satisfy the time-regular and frequency-regular properties are identified as the regularity of daily behaviors. Then, the irregularity probability of the daily behaviors in one days can be calculated based on the selected regular behaviors. Experiments show that the proposed DBID has a good performance in terms of F measure, compared the existing mechanisms.
{"title":"Irregularity Detection of Daily Behavior Pattern Based on Regularity Feature Extraction for Home Elderly","authors":"Cuijuan Shang, Chih-Yung Chang, Qiaoyun Zhang, Shih-Jung Wu","doi":"10.1109/ICCE-Taiwan55306.2022.9869221","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869221","url":null,"abstract":"Daily behavior irregularity detection is important for assessment of the health status for the elderly in homecare. This paper proposes a Daily Behavior Irregularity Detection (DBID) mechanism which outputs the irregularity probability of daily behaviors based on the extracted regularity features using unsupervised learning algorithm. The regular behaviors which satisfy the time-regular and frequency-regular properties are identified as the regularity of daily behaviors. Then, the irregularity probability of the daily behaviors in one days can be calculated based on the selected regular behaviors. Experiments show that the proposed DBID has a good performance in terms of F measure, compared the existing mechanisms.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127146609","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869191
Jing-Ming Guo, S. Sankarasrinivasan
Error diffusion halftoning is one of the widely adopted technique in printers, to transform the gray-scale image into its approximate binary version. Further, the classification of halftones is very important to facilitate inverse halftoning, source printer identification, forensics analysis and other halftone processing tasks. Practically, majority of the printed documents are unlabeled and hence hard to train using supervised approach. This study exploits the advantage of self-supervised learning (SSL), in particular the simplified framework for contrastive learning of visual representation in learning best representation features for halftones. As the data augmentation play a critical role in SSL models, and this study focus on optimization of the existing augmentations and also added new random augmentation techniques to enhance the feature learning. In addition, different variants of ResNet backbone is tried to find the ideal case, and the error diffusion dataset is also generated for analysis. From detailed experiments, it has been found that the proposed method can perform consistent with supervised learning approach without large labelled data.
{"title":"Error Diffusion Halftone Classification using Contrastive Learning","authors":"Jing-Ming Guo, S. Sankarasrinivasan","doi":"10.1109/ICCE-Taiwan55306.2022.9869191","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869191","url":null,"abstract":"Error diffusion halftoning is one of the widely adopted technique in printers, to transform the gray-scale image into its approximate binary version. Further, the classification of halftones is very important to facilitate inverse halftoning, source printer identification, forensics analysis and other halftone processing tasks. Practically, majority of the printed documents are unlabeled and hence hard to train using supervised approach. This study exploits the advantage of self-supervised learning (SSL), in particular the simplified framework for contrastive learning of visual representation in learning best representation features for halftones. As the data augmentation play a critical role in SSL models, and this study focus on optimization of the existing augmentations and also added new random augmentation techniques to enhance the feature learning. In addition, different variants of ResNet backbone is tried to find the ideal case, and the error diffusion dataset is also generated for analysis. From detailed experiments, it has been found that the proposed method can perform consistent with supervised learning approach without large labelled data.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"159 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126027375","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869280
Chin-Fa Hsieh, Li‐Chi Chen, Zhe-Hao Lin
The Bresenham's line algorithm is one used to draw a straight line determined by two points, which is often implemented in graphics chips. In order to speed up the drawing performance, the hardware circuit in this paper is designed by an FPGA based on its advantages of synchronous execution and multiplexing of multiple circuits. This FPGA hardware circuit is verified with a three-axis motion controller platform which is composed of three two-phase stepping motors and three rotary screws. The Bresenham's line algorithm, to calculate the coordinates of the motion path of circles by using five groups of coordinate-operation units, is implemented in this work. The coordinates calculated by VerilogHDL and the coordinates of the motion path calculated by C language are compared to verify the correctness, and then the graph is drawn to present the results. The experimental results show that, by using one group of coordinate operation units as in the tradition and by using five groups of coordinate-operation units in this paper, their execution cycles turn out to be 21 and 6, respectively. Obviously, this work can achieve to accelerate the function of drawing.
{"title":"High Performance Architecture of a Graphics Accelerator","authors":"Chin-Fa Hsieh, Li‐Chi Chen, Zhe-Hao Lin","doi":"10.1109/ICCE-Taiwan55306.2022.9869280","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869280","url":null,"abstract":"The Bresenham's line algorithm is one used to draw a straight line determined by two points, which is often implemented in graphics chips. In order to speed up the drawing performance, the hardware circuit in this paper is designed by an FPGA based on its advantages of synchronous execution and multiplexing of multiple circuits. This FPGA hardware circuit is verified with a three-axis motion controller platform which is composed of three two-phase stepping motors and three rotary screws. The Bresenham's line algorithm, to calculate the coordinates of the motion path of circles by using five groups of coordinate-operation units, is implemented in this work. The coordinates calculated by VerilogHDL and the coordinates of the motion path calculated by C language are compared to verify the correctness, and then the graph is drawn to present the results. The experimental results show that, by using one group of coordinate operation units as in the tradition and by using five groups of coordinate-operation units in this paper, their execution cycles turn out to be 21 and 6, respectively. Obviously, this work can achieve to accelerate the function of drawing.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125543605","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9868974
Hsiao-Ting Tseng, Y. Zheng, Chen-Chiung Hsieh
Due to the epidemic situation, in order to greatly reduce the infection risk of face-to-face interviews, this paper implements the BERT combined with RCNN to judge the positive and negative directions of the text, and then uses BERT's next sentence prediction (NSP) to find out the topic-related sentences in the text. Finally, a cognitive dictionary is used to calculate the degree of agreement or disagreement, so as to obtain the degree of support of the reviewer. This paper is also useful for letting visitors or authors know what the respondents' views are.
{"title":"Sentiment Analysis using BERT, LSTM, and Cognitive Dictionary","authors":"Hsiao-Ting Tseng, Y. Zheng, Chen-Chiung Hsieh","doi":"10.1109/ICCE-Taiwan55306.2022.9868974","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9868974","url":null,"abstract":"Due to the epidemic situation, in order to greatly reduce the infection risk of face-to-face interviews, this paper implements the BERT combined with RCNN to judge the positive and negative directions of the text, and then uses BERT's next sentence prediction (NSP) to find out the topic-related sentences in the text. Finally, a cognitive dictionary is used to calculate the degree of agreement or disagreement, so as to obtain the degree of support of the reviewer. This paper is also useful for letting visitors or authors know what the respondents' views are.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116539929","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869014
QinXin Zhan, Xin-Zhu Li, Xin Kang, Shau-Yu Lu
Commercially drone are mainly operated by dedicated personnel with trained skills. The high learning cost of operation will discourage some potential users. In this paper, we propose an human-machine Interaction to drone by intercepting the visual images of robots and use Generative Adversarial Network(GAN) to train. A camera is used to intercept the operator's gestures, and the photos with the operator's gestures are converted into control commands to improve the accuracy of operation in complex backgrounds. As a result of this research, drone flight control can be accomplished in more complex backgrounds, greatly simplifying operator stress.
{"title":"A Drone Human-Machine Interaction Method Based on Generative Adversarial Network","authors":"QinXin Zhan, Xin-Zhu Li, Xin Kang, Shau-Yu Lu","doi":"10.1109/ICCE-Taiwan55306.2022.9869014","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869014","url":null,"abstract":"Commercially drone are mainly operated by dedicated personnel with trained skills. The high learning cost of operation will discourage some potential users. In this paper, we propose an human-machine Interaction to drone by intercepting the visual images of robots and use Generative Adversarial Network(GAN) to train. A camera is used to intercept the operator's gestures, and the photos with the operator's gestures are converted into control commands to improve the accuracy of operation in complex backgrounds. As a result of this research, drone flight control can be accomplished in more complex backgrounds, greatly simplifying operator stress.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122418012","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}
The main purpose of this paper is to develop a smart recycling bin to replace the manual sorting of waste. This paper extends the previous work by 1) adding an ultrasonic module to assess the amount of waste in the recycling bin, 2) distinguishing one more class, i.e. the glass bottles, in addition to metal cans, plastic bottles, and Tetra Pak cartons, and 3) developing a mobile App to real-time monitor the status of recycling bins. Experimental results show that the developed system has the over 96% mean average precision (mAP) for the Tetra Pak cartons and plastic bottles.
{"title":"An Extended Smart Recycling Bins Using Deep Learning Networks","authors":"Jin-Shyan Lee, Jui-Wen Chen, Jia-Chen Lai, Gin-Lin Huang","doi":"10.1109/ICCE-Taiwan55306.2022.9868985","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9868985","url":null,"abstract":"The main purpose of this paper is to develop a smart recycling bin to replace the manual sorting of waste. This paper extends the previous work by 1) adding an ultrasonic module to assess the amount of waste in the recycling bin, 2) distinguishing one more class, i.e. the glass bottles, in addition to metal cans, plastic bottles, and Tetra Pak cartons, and 3) developing a mobile App to real-time monitor the status of recycling bins. Experimental results show that the developed system has the over 96% mean average precision (mAP) for the Tetra Pak cartons and plastic bottles.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122767036","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 : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869039
Yu-Chen Su, Tsung-Jung Liu, Kuan-Hsien Liu
In recent development of semantic segmentation, the deep convolutional encoder-decoder has become mainstream schemes for remote sensing images. In this paper, we proposed a U-Net like architecture for segmentation of remote sensing images using wavelet frequency channel attention (WFCA) blocks.
{"title":"Wavelet Frequency Channel Attention on Remote Sensing Image Segmentation","authors":"Yu-Chen Su, Tsung-Jung Liu, Kuan-Hsien Liu","doi":"10.1109/ICCE-Taiwan55306.2022.9869039","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869039","url":null,"abstract":"In recent development of semantic segmentation, the deep convolutional encoder-decoder has become mainstream schemes for remote sensing images. In this paper, we proposed a U-Net like architecture for segmentation of remote sensing images using wavelet frequency channel attention (WFCA) blocks.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122614942","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：