Pub Date : 2022-07-06DOI: 10.1109/ICCE-Taiwan55306.2022.9869187
Po-Han Chen, Sin-Ye Jhong, Chih-Hsien Hsia
As the global demand for coffee rises, coffee has become a part of the daily lives of many. The taste of the brewed coffee is closely related to the quality of coffee beans, which has led to many researchers developing automated methods to accurately distinguish good coffee beans from bad ones. The research often used supervised learning technology by utilizing large sets of labeled data for training, but the labeling requires a substantial amount of manpower that is impractical for real production line usage. To solve this problem, we proposed a method that the combines semi-supervised learning and attention mechanism to classify the two types of coffee beans. Through explainable consistency training and directional attention algorithm, we solve the high-cost problem of labeling data and strengthen the prediction ability of the model. The experimental results show that the study has high classification performance and can achieve an F1-score of 97.21%.
{"title":"Semi-Supervised Learning with Attention-Based CNN for Classification of Coffee Beans Defect","authors":"Po-Han Chen, Sin-Ye Jhong, Chih-Hsien Hsia","doi":"10.1109/ICCE-Taiwan55306.2022.9869187","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869187","url":null,"abstract":"As the global demand for coffee rises, coffee has become a part of the daily lives of many. The taste of the brewed coffee is closely related to the quality of coffee beans, which has led to many researchers developing automated methods to accurately distinguish good coffee beans from bad ones. The research often used supervised learning technology by utilizing large sets of labeled data for training, but the labeling requires a substantial amount of manpower that is impractical for real production line usage. To solve this problem, we proposed a method that the combines semi-supervised learning and attention mechanism to classify the two types of coffee beans. Through explainable consistency training and directional attention algorithm, we solve the high-cost problem of labeling data and strengthen the prediction ability of the model. The experimental results show that the study has high classification performance and can achieve an F1-score of 97.21%.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"91 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":"132980223","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-Taiwan55306.2022.9869163
Yi-Fan Wu, Chian-Huey Liaw, Yu-Hsuan Lee
Image dehazing is an image processing technique to restore a hazy image back to hazy-free one. Airlight estimation plays an important role in image dehazing algorithm. Dark Channel Prior (DCP) is an efficient algorithm to predict airlight. However, the sorting process of DCP causes tremendous computation requirement, limiting its potential in image dehazing chip design. To overcome this limitation, Down-sampling DCP (DS-DCP) is proposed to provide a low complexity algorithm for airlight estimation. Experiment results demonstrate that the computation saving ratio (CSR) of DS-DCP is as high as 98%, while keeping error as minor as 0.22%.
{"title":"Down-Sampling Dark Channel Prior of Airlight Estimation for Low Complexity Image Dehazing Chip Design","authors":"Yi-Fan Wu, Chian-Huey Liaw, Yu-Hsuan Lee","doi":"10.1109/ICCE-Taiwan55306.2022.9869163","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869163","url":null,"abstract":"Image dehazing is an image processing technique to restore a hazy image back to hazy-free one. Airlight estimation plays an important role in image dehazing algorithm. Dark Channel Prior (DCP) is an efficient algorithm to predict airlight. However, the sorting process of DCP causes tremendous computation requirement, limiting its potential in image dehazing chip design. To overcome this limitation, Down-sampling DCP (DS-DCP) is proposed to provide a low complexity algorithm for airlight estimation. Experiment results demonstrate that the computation saving ratio (CSR) of DS-DCP is as high as 98%, while keeping error as minor as 0.22%.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"236 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113982662","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-Taiwan55306.2022.9869234
Xin-Yu Shih, Hsiang-En Wu
In this paper, we propose a design methodology of queue-based fast classification for sequential minimal optimization (SMO) in support vector machine (SVM) training. The queue is designed to tremendously reduce the searching space of weightings. Our method is useful to simplify operating steps of SMO and almost achieve the same performance in terms of classification accuracy with respect to full-search approach. In the Matlab simulation, our method is completely verified with 6 representative data sets. As compared to full-search and heuristic approaches, the running speed of our method is increased by 7.53 and 2.91 times, respectively. It features high efficiency without sacrificing classification accuracy.
{"title":"Design Methodology of Queue-Based Fast Classification for Sequential Minimal Optimization in SVM ML-Training","authors":"Xin-Yu Shih, Hsiang-En Wu","doi":"10.1109/ICCE-Taiwan55306.2022.9869234","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869234","url":null,"abstract":"In this paper, we propose a design methodology of queue-based fast classification for sequential minimal optimization (SMO) in support vector machine (SVM) training. The queue is designed to tremendously reduce the searching space of weightings. Our method is useful to simplify operating steps of SMO and almost achieve the same performance in terms of classification accuracy with respect to full-search approach. In the Matlab simulation, our method is completely verified with 6 representative data sets. As compared to full-search and heuristic approaches, the running speed of our method is increased by 7.53 and 2.91 times, respectively. It features high efficiency without sacrificing classification accuracy.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"232 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":"122465599","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-Taiwan55306.2022.9869149
Bo-Xi Lai, Shih-Hsu Huang, Hsu-Yu Kao
Computing-in-memories (CIM) is recognized as a useful design technique for eliminating the Von Neumann bottleneck. However, there is a need for circuit designers to determine the configuration (i.e., design parameters) of CIM-based accelerators. Note that the configuration has influences on circuit area, throughput, and energy efficiency. In this paper, we focus on the SRAM CIM-based accelerator design. A reinforcement learning methodology is proposed to assist circuit designers to find the most suitable configuration. Experiment data show that the proposed methodology works well in practice.
{"title":"A Reinforcement Learning Methodology for The Search of SRAM CIM-based Accelerator Configuration","authors":"Bo-Xi Lai, Shih-Hsu Huang, Hsu-Yu Kao","doi":"10.1109/ICCE-Taiwan55306.2022.9869149","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869149","url":null,"abstract":"Computing-in-memories (CIM) is recognized as a useful design technique for eliminating the Von Neumann bottleneck. However, there is a need for circuit designers to determine the configuration (i.e., design parameters) of CIM-based accelerators. Note that the configuration has influences on circuit area, throughput, and energy efficiency. In this paper, we focus on the SRAM CIM-based accelerator design. A reinforcement learning methodology is proposed to assist circuit designers to find the most suitable configuration. Experiment data show that the proposed methodology works well in practice.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"42 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":"123886647","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}
Enhancing the immersive experience in virtual reality (VR) through external force has always been a hot topic. The development and research of external devices have been published from time to time. Various interactive devices could increase the sensory experience at the tactile aspect in addition to sight and hearing. This research has proposed a device for VR, which could adjust its rotational speed through the individual flywheels on six axes of tilt to generate changes in inertia moment to create force feedback. The device mainly acts on the user's hand to simulate the feedback felt by the hand in the VR world: 1) the reaction force and friction of the handheld weapon hitting the object, 2) the fictitious magic effect, 3) simulating different environments to feel the force feedback brought by gravity and resistance to the hand, to achieve a more substantial experience immersion.
{"title":"Blooming: A Handheld Device Using Flywheel to Simulates Various Multi-Force Feedback","authors":"Zi-Han Xu, Hung-Ju Wei, Tzu-Hsuan Yeh, Chi Fang, Ju-Chun Ko, Saiau-Yue Tsau, Ko-Chiu Wu","doi":"10.1109/ICCE-Taiwan55306.2022.9869208","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869208","url":null,"abstract":"Enhancing the immersive experience in virtual reality (VR) through external force has always been a hot topic. The development and research of external devices have been published from time to time. Various interactive devices could increase the sensory experience at the tactile aspect in addition to sight and hearing. This research has proposed a device for VR, which could adjust its rotational speed through the individual flywheels on six axes of tilt to generate changes in inertia moment to create force feedback. The device mainly acts on the user's hand to simulate the feedback felt by the hand in the VR world: 1) the reaction force and friction of the handheld weapon hitting the object, 2) the fictitious magic effect, 3) simulating different environments to feel the force feedback brought by gravity and resistance to the hand, to achieve a more substantial experience immersion.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"34 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":"123939925","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-Taiwan55306.2022.9869017
J. Liou, Jia-Wei Jiang
In this research, it is connected with the chinese medicine treatment, and the preliminary ideas are proposed, such as acupuncture and electrotherapy. By applying voltage to the skin tissues on both sides of the human meridian, the current value of the meridian circuit is detected to complete the interpretation of the human meridian information.
{"title":"The Observation of Physiological Signals with Electroacupuncture Stimulation and Preliminary Ideas","authors":"J. Liou, Jia-Wei Jiang","doi":"10.1109/ICCE-Taiwan55306.2022.9869017","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869017","url":null,"abstract":"In this research, it is connected with the chinese medicine treatment, and the preliminary ideas are proposed, such as acupuncture and electrotherapy. By applying voltage to the skin tissues on both sides of the human meridian, the current value of the meridian circuit is detected to complete the interpretation of the human meridian information.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"3 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":"127847628","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-Taiwan55306.2022.9869245
Kenta Sawada, Kazuhisa Nakasho, K. Wasaki, N. Shimoi
In this paper, we propose a method for classifying human postures using RFID and machine learning, and discuss the learning accuracy of this method.
本文提出了一种基于RFID和机器学习的人体姿势分类方法，并讨论了该方法的学习精度。
{"title":"Classification of Human Posture on Bed Using Machine Learning","authors":"Kenta Sawada, Kazuhisa Nakasho, K. Wasaki, N. Shimoi","doi":"10.1109/ICCE-Taiwan55306.2022.9869245","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869245","url":null,"abstract":"In this paper, we propose a method for classifying human postures using RFID and machine learning, and discuss the learning accuracy of this method.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"33 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":"121324860","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-Taiwan55306.2022.9869135
C. Tseng, Su-Ling Lee
In this paper, a temperature data denoising method using directed Laplacian matrix (DLM) and heat kernel smoothing (HKS) is presented. First, the temperature data collected from sensor network is represented as the directed graph signal. Then, the adjacency matrix and degree matrix of directed graph is used to define the DLM. And, directed graph Fourier transform is defined by the eigen-decomposition of DLM. Next, the HKS filter is employed to reduce the noise superimposed on the temperature data. Using the Taylor series expansion, the HKS filter can be approximated by a polynomial digraph filter to get a distributed implementation in vertex domain. Finally, the performance of proposed denoising method is evaluated by the real-word temperature data to show its effectiveness.
{"title":"Temperature Data Denoising Based on Directed Laplacian Matrix and Heat Kernel Smoothing","authors":"C. Tseng, Su-Ling Lee","doi":"10.1109/ICCE-Taiwan55306.2022.9869135","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869135","url":null,"abstract":"In this paper, a temperature data denoising method using directed Laplacian matrix (DLM) and heat kernel smoothing (HKS) is presented. First, the temperature data collected from sensor network is represented as the directed graph signal. Then, the adjacency matrix and degree matrix of directed graph is used to define the DLM. And, directed graph Fourier transform is defined by the eigen-decomposition of DLM. Next, the HKS filter is employed to reduce the noise superimposed on the temperature data. Using the Taylor series expansion, the HKS filter can be approximated by a polynomial digraph filter to get a distributed implementation in vertex domain. Finally, the performance of proposed denoising method is evaluated by the real-word temperature data to show its effectiveness.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"7 12 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":"128423256","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}
This paper presents a cost-effective video stabilization method for fast and large-shaking frames. To achieve real-time and high-quality video stabilization for fast and large-shaking frames, the main strategy of the proposed method is to find the optimal feature-point match to generate the better transformation matrix. Besides, the image pre-processing is exploited to down-sample the picture and then set the ROI area for substantially increasing the speed of the subsequent processing without affecting the detection of feature points. The proposed method is more cost-effective in stabilization than other approaches, especially for fast and large-shaking frames (e.g., video-shooting while running) and can be applied to the wearable cameras, sports cameras, and vehicle cameras.
{"title":"The Cost-Effective Video Stabilization Method for Wearable Camera","authors":"Chao-Ho Chen, Chia-En Lin, Tsong-Yi Chen, Da-Jinn Wang, Cheng-Fu Liao, Cheng-Kang Wen","doi":"10.1109/ICCE-Taiwan55306.2022.9869216","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869216","url":null,"abstract":"This paper presents a cost-effective video stabilization method for fast and large-shaking frames. To achieve real-time and high-quality video stabilization for fast and large-shaking frames, the main strategy of the proposed method is to find the optimal feature-point match to generate the better transformation matrix. Besides, the image pre-processing is exploited to down-sample the picture and then set the ROI area for substantially increasing the speed of the subsequent processing without affecting the detection of feature points. The proposed method is more cost-effective in stabilization than other approaches, especially for fast and large-shaking frames (e.g., video-shooting while running) and can be applied to the wearable cameras, sports cameras, and vehicle cameras.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"100 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":"116653980","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-Taiwan55306.2022.9869101
Chih-Jer Lin, Ting–Yun Chiu
According to Taiwan Printed Circuit Association (TPCA) statistics from 2017 to 2021, the output value of PCBs in Taiwan has increased year by year or even broken through new highs, and the cost of labor and time to visually inspect scratches on PABA by personnel has increased relatively. Therefore, this study focuses on PCBA for scratch detection and trains multiple models based on semantic segmentation UNET network architecture. The proposed D-LinkNet is optimized to reduce the problem of missed detection and misclassification caused by complex backgrounds and long span of defects. By comparing various attention modules in different positions and types to improve the accuracy, and using the dilated convolution instead of pooling layer, the encoder-decoder structure is optimized to reduce the loss of information in the downsampling process, simultaneously improve attention module effect. In addition, this experiment uses a small amount of data to increase the amount of data by cutting and augmenting the data, and compares the effect of image cutting size on the accuracy rate to find the best data size for training, and uses IoU as the model scoring method to apply the model with the best segmentation effect to more scratch detection tasks and reduce the labor cost at the factory.
{"title":"Optimize D-LinkNet for Printed Circuit Board Defects Inspection","authors":"Chih-Jer Lin, Ting–Yun Chiu","doi":"10.1109/ICCE-Taiwan55306.2022.9869101","DOIUrl":"https://doi.org/10.1109/ICCE-Taiwan55306.2022.9869101","url":null,"abstract":"According to Taiwan Printed Circuit Association (TPCA) statistics from 2017 to 2021, the output value of PCBs in Taiwan has increased year by year or even broken through new highs, and the cost of labor and time to visually inspect scratches on PABA by personnel has increased relatively. Therefore, this study focuses on PCBA for scratch detection and trains multiple models based on semantic segmentation UNET network architecture. The proposed D-LinkNet is optimized to reduce the problem of missed detection and misclassification caused by complex backgrounds and long span of defects. By comparing various attention modules in different positions and types to improve the accuracy, and using the dilated convolution instead of pooling layer, the encoder-decoder structure is optimized to reduce the loss of information in the downsampling process, simultaneously improve attention module effect. In addition, this experiment uses a small amount of data to increase the amount of data by cutting and augmenting the data, and compares the effect of image cutting size on the accuracy rate to find the best data size for training, and uses IoU as the model scoring method to apply the model with the best segmentation effect to more scratch detection tasks and reduce the labor cost at the factory.","PeriodicalId":164671,"journal":{"name":"2022 IEEE International Conference on Consumer Electronics - Taiwan","volume":"47 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":"117334790","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：