Pub Date : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937129
Jie-Min Lin, Wei-Liang Lin, Chih-Peng Fan
Based on the information of body proportion, in this study, a simple and effective processing scheme is developed for two age groups classification, i.e. children and adults for the applications of smart autonomous movers. By the YOLO-based CNN model for head and body objects detections, the recognition accuracies of age group classification for children and adults are 95% and 92.5% respectively with the image datasets collected in publics. Compared with the existed design, the proposed methodology performs simpler and more effective recognition capability for age group classification of adults and children. The proposed design is implemented on GPU-based embedded platform for real-time applications.
{"title":"Age Group Classifier of Adults and Children with YOLO-based Deep Learning Pre-Processing Scheme for Embedded Platforms","authors":"Jie-Min Lin, Wei-Liang Lin, Chih-Peng Fan","doi":"10.1109/ICCE-Berlin56473.2022.9937129","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937129","url":null,"abstract":"Based on the information of body proportion, in this study, a simple and effective processing scheme is developed for two age groups classification, i.e. children and adults for the applications of smart autonomous movers. By the YOLO-based CNN model for head and body objects detections, the recognition accuracies of age group classification for children and adults are 95% and 92.5% respectively with the image datasets collected in publics. Compared with the existed design, the proposed methodology performs simpler and more effective recognition capability for age group classification of adults and children. The proposed design is implemented on GPU-based embedded platform for real-time applications.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125665835","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937123
Arne Aarts, Wil Michiels, Peter Roelse
Object detectors based on neural networks are deployed in various consumer electronics products to predict different types of object and their location in images. This paper presents a cloning attack on object detectors, using problem domain samples and oracle access to a trained object detector. As in known cloning attacks on image classifiers, the presented attack uses the oracle access to label the samples. The resulting set of labeled samples, referred to as the surrogate dataset, is then used to train the clone detector. Compared to image classifiers, the surrogate dataset created by an object detector can contain more types of error. The paper describes a way to assess the quality of the surrogate dataset. The cloning attack was implemented, and experiments were conducted with a CenterNet and a RetinaNet object detector, and the Oxford-IIIT Pet, Tsinghua-Tencent 100K, and WIDER FACE datasets. The results show that object detectors can be cloned successfully, even if the quality of the surrogate dataset is relatively low. However, in case of a low-quality surrogate dataset, the quality of the clone detector was only high if it used the same architecture as the target detector.
{"title":"Cloning Object Detectors","authors":"Arne Aarts, Wil Michiels, Peter Roelse","doi":"10.1109/ICCE-Berlin56473.2022.9937123","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937123","url":null,"abstract":"Object detectors based on neural networks are deployed in various consumer electronics products to predict different types of object and their location in images. This paper presents a cloning attack on object detectors, using problem domain samples and oracle access to a trained object detector. As in known cloning attacks on image classifiers, the presented attack uses the oracle access to label the samples. The resulting set of labeled samples, referred to as the surrogate dataset, is then used to train the clone detector. Compared to image classifiers, the surrogate dataset created by an object detector can contain more types of error. The paper describes a way to assess the quality of the surrogate dataset. The cloning attack was implemented, and experiments were conducted with a CenterNet and a RetinaNet object detector, and the Oxford-IIIT Pet, Tsinghua-Tencent 100K, and WIDER FACE datasets. The results show that object detectors can be cloned successfully, even if the quality of the surrogate dataset is relatively low. However, in case of a low-quality surrogate dataset, the quality of the clone detector was only high if it used the same architecture as the target detector.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127252950","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937109
Kiavash Fathi, Alireza Darvishy, H. W. V. D. Venn
With the Augmented Reality (AR) technology avail-able today, it is quite feasible to accommodate the needs of the visually impaired (VI) via AR. In this paper, a framework is introduced to help the VI navigate and explore unfamiliar indoor environments. In contrast to commonly used AR applications focused on visual augmentation, the proposed framework em-ploys auditory three-dimensional feedback (A3DF) for guiding the VI. Concretely, the current framework reads the pose of the user and helps the VI reach a target location via A3DF. The A3DF is implemented with the Unity game engine to provide the optimal user experience. After acquiring the environment mesh (EM), the optimal path from the user's location to the target location is calculated, while avoiding obstacles using Unity's navigation system. Moreover, the user is provided with semantic information about the unknown environment whilst exploring via auditory information. This framework is implemented on Microsoft HoloLens 2 and tested at an office environment with different locations of interest. Additionally, this framework potentially accelerates the learning curve since the user can be trained using Unity's simulation environment. Lastly, given different design parameters of the framework, the proposed method can be fine-tuned to fit the specific needs of the individual VI.
{"title":"Augmented Reality for the Visually Impaired: Navigation Aid and Scene Semantics for Indoor Use Cases","authors":"Kiavash Fathi, Alireza Darvishy, H. W. V. D. Venn","doi":"10.1109/ICCE-Berlin56473.2022.9937109","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937109","url":null,"abstract":"With the Augmented Reality (AR) technology avail-able today, it is quite feasible to accommodate the needs of the visually impaired (VI) via AR. In this paper, a framework is introduced to help the VI navigate and explore unfamiliar indoor environments. In contrast to commonly used AR applications focused on visual augmentation, the proposed framework em-ploys auditory three-dimensional feedback (A3DF) for guiding the VI. Concretely, the current framework reads the pose of the user and helps the VI reach a target location via A3DF. The A3DF is implemented with the Unity game engine to provide the optimal user experience. After acquiring the environment mesh (EM), the optimal path from the user's location to the target location is calculated, while avoiding obstacles using Unity's navigation system. Moreover, the user is provided with semantic information about the unknown environment whilst exploring via auditory information. This framework is implemented on Microsoft HoloLens 2 and tested at an office environment with different locations of interest. Additionally, this framework potentially accelerates the learning curve since the user can be trained using Unity's simulation environment. Lastly, given different design parameters of the framework, the proposed method can be fine-tuned to fit the specific needs of the individual VI.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121097297","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937114
Sujitha Venkatapathy, In-ho Ra, Han-Gue Jo
Network virtualization (NV) provides a feasible mechanism for operating numerous diverse virtual networks concurrently on a shared physical infrastructure network. The key issue in NV is virtual network embedding (VNE), which efficiently and effectively maps virtualized networks (VNs) with multiple resource needs for nodes and links to the underlying physical network with limited resources. A multiple topological attributes (MTA) based embedding algorithm is proposed to address the issue of providing different virtual request ser-vices delivered in a wireless network environment, leading to an unstable utilization of physical network resources and a low access rate for subsequent requests. It is emphasized that machine learning (ML) should be integrated into the process of network slicing in order to properly classify the received wireless virtual request. In this work, virtual request services are categorized automatically using support vector machine (SVM), and resources are allocated accordingly. The proposed technique organizes nodes in the embedding process according to their priority based on multiple topological properties of virtual and physical networks. According to the findings of the simulations, the SVM-MTA algorithm enhances both the acceptance rate and the resource efficiency of the network.
{"title":"A Self-Adaptive Wireless Network Service Embedding through SVM and MTA","authors":"Sujitha Venkatapathy, In-ho Ra, Han-Gue Jo","doi":"10.1109/ICCE-Berlin56473.2022.9937114","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937114","url":null,"abstract":"Network virtualization (NV) provides a feasible mechanism for operating numerous diverse virtual networks concurrently on a shared physical infrastructure network. The key issue in NV is virtual network embedding (VNE), which efficiently and effectively maps virtualized networks (VNs) with multiple resource needs for nodes and links to the underlying physical network with limited resources. A multiple topological attributes (MTA) based embedding algorithm is proposed to address the issue of providing different virtual request ser-vices delivered in a wireless network environment, leading to an unstable utilization of physical network resources and a low access rate for subsequent requests. It is emphasized that machine learning (ML) should be integrated into the process of network slicing in order to properly classify the received wireless virtual request. In this work, virtual request services are categorized automatically using support vector machine (SVM), and resources are allocated accordingly. The proposed technique organizes nodes in the embedding process according to their priority based on multiple topological properties of virtual and physical networks. According to the findings of the simulations, the SVM-MTA algorithm enhances both the acceptance rate and the resource efficiency of the network.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121464175","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937133
S. Bianco, M. Buzzelli
In this paper we propose the truncated edge-based color constancy. It is based on, and extends, the edge-based framework by introducing the use of truncated Gaussian filters. The truncation level can be controlled with the use of a dedicated parameter that is added to the other three parameters existing in the edge-based framework, namely the derivative order, the standard deviation of the Gaussian filter, and the Minkowski norm. Experimental results on two standard dataset for color constancy show that the truncated edge-based framework allows to achieve the same or higher illuminant estimation accuracy of the edge-based framework considerably reducing the number of operations.
{"title":"Truncated Edge-based Color Constancy","authors":"S. Bianco, M. Buzzelli","doi":"10.1109/ICCE-Berlin56473.2022.9937133","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937133","url":null,"abstract":"In this paper we propose the truncated edge-based color constancy. It is based on, and extends, the edge-based framework by introducing the use of truncated Gaussian filters. The truncation level can be controlled with the use of a dedicated parameter that is added to the other three parameters existing in the edge-based framework, namely the derivative order, the standard deviation of the Gaussian filter, and the Minkowski norm. Experimental results on two standard dataset for color constancy show that the truncated edge-based framework allows to achieve the same or higher illuminant estimation accuracy of the edge-based framework considerably reducing the number of operations.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115920362","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937125
Antonia Antoniello, Antonio Sabatelli, Simone Valenti, Maria Di Tillo, L. Pepa, L. Spalazzi, E. Andrenelli, M. Capecci, M. Ceravolo
This article presents the architecture of a telerehabilitation and telemonitoring system for people with Parkinson's disease. The main advantages of the system are the use of low- cost and widespread consumer technology devices. A pilot study on 5 patients allowed a first assessment of the technical reliability, usability, and acceptability of the system, helping the technical and clinical staff to improve the system.
{"title":"A low-cost telerehabilitation and telemonitoring system for people with Parkinson's disease: the architecture","authors":"Antonia Antoniello, Antonio Sabatelli, Simone Valenti, Maria Di Tillo, L. Pepa, L. Spalazzi, E. Andrenelli, M. Capecci, M. Ceravolo","doi":"10.1109/ICCE-Berlin56473.2022.9937125","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937125","url":null,"abstract":"This article presents the architecture of a telerehabilitation and telemonitoring system for people with Parkinson's disease. The main advantages of the system are the use of low- cost and widespread consumer technology devices. A pilot study on 5 patients allowed a first assessment of the technical reliability, usability, and acceptability of the system, helping the technical and clinical staff to improve the system.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123723579","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937112
Seung-Bum Lee, Lukasz Dudek, Piotr Wojdyna
This paper presents FastShare, a novel high performance push-based file sharing approach on wireless closes-range multiple device environment. In order to enhance previous approach, FileShare, FastShare includes TCP connection reduction, push-based approach and notification simplification. Compared with previous and existing approaches, FastShare shows excellent performance with multiple small files and even fair results with large sized files in terms of delivery.
{"title":"FastShare: push-based file sharing approach on wireless multi device environment","authors":"Seung-Bum Lee, Lukasz Dudek, Piotr Wojdyna","doi":"10.1109/ICCE-Berlin56473.2022.9937112","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937112","url":null,"abstract":"This paper presents FastShare, a novel high performance push-based file sharing approach on wireless closes-range multiple device environment. In order to enhance previous approach, FileShare, FastShare includes TCP connection reduction, push-based approach and notification simplification. Compared with previous and existing approaches, FastShare shows excellent performance with multiple small files and even fair results with large sized files in terms of delivery.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"35 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133565545","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937117
S. K. Datta
The paper introduces a novel secure, interoperable, and end-to-end industry 4.0 service platform for lot-size-one man-ufacturing. The challenges faced by the European manufacturing industry that prevent it from embracing such a new paradigm are outlined. Then, the platform architecture and operational steps are summarised.
{"title":"Secure, Interoperable, End-to-End Industry 4.0 Service Platform for Lot-Size-One Manufacturing","authors":"S. K. Datta","doi":"10.1109/ICCE-Berlin56473.2022.9937117","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937117","url":null,"abstract":"The paper introduces a novel secure, interoperable, and end-to-end industry 4.0 service platform for lot-size-one man-ufacturing. The challenges faced by the European manufacturing industry that prevent it from embracing such a new paradigm are outlined. Then, the platform architecture and operational steps are summarised.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121483115","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 : 2022-09-02DOI: 10.1109/ICCE-Berlin56473.2022.9937110
Po-Yung Chou, Cheng-Hung Lin, W. Kao, Yi-Fang Lee, Chen-Chien James Hsu
Deep learning has developed rapidly in recent years, not only in image recognition, but now also in action recognition. The research on action recognition started with 3D-CNN, which has achieved good results on many tasks. But most action recognition networks have room for improvement in fine-grained action recognition. The reason is that there is only a slight difference between categories in the fine-grained classification task. e.g. basketball fouls only occur in a few frames and a small region. This situation may lead to some errors with 3D-CNN methods because these models tend to merge all temporal features. To identify these fouls, it is necessary to strengthen the detection of small periods. In this paper, we propose a temporal score network suitable for existing networks, including 3D-Resnet50, 3D-wide-Resnet50, $mathbf{R}mathbf{(}mathbf{2}mathbf{+}mathbf{1}mathbf{)}$ D-Resnet50, and I3D-50 to improve the accuracy of fine-grained action recognition. The experimental results show that the accuracy of various models is improved by 3.85% to 6% after adding the proposed network. Since there is no relevant public dataset, we collect the data ourselves to create a basketball foul dataset.
{"title":"A Temporal Scores Network for Basketball Foul Classification","authors":"Po-Yung Chou, Cheng-Hung Lin, W. Kao, Yi-Fang Lee, Chen-Chien James Hsu","doi":"10.1109/ICCE-Berlin56473.2022.9937110","DOIUrl":"https://doi.org/10.1109/ICCE-Berlin56473.2022.9937110","url":null,"abstract":"Deep learning has developed rapidly in recent years, not only in image recognition, but now also in action recognition. The research on action recognition started with 3D-CNN, which has achieved good results on many tasks. But most action recognition networks have room for improvement in fine-grained action recognition. The reason is that there is only a slight difference between categories in the fine-grained classification task. e.g. basketball fouls only occur in a few frames and a small region. This situation may lead to some errors with 3D-CNN methods because these models tend to merge all temporal features. To identify these fouls, it is necessary to strengthen the detection of small periods. In this paper, we propose a temporal score network suitable for existing networks, including 3D-Resnet50, 3D-wide-Resnet50, $mathbf{R}mathbf{(}mathbf{2}mathbf{+}mathbf{1}mathbf{)}$ D-Resnet50, and I3D-50 to improve the accuracy of fine-grained action recognition. The experimental results show that the accuracy of various models is improved by 3.85% to 6% after adding the proposed network. Since there is no relevant public dataset, we collect the data ourselves to create a basketball foul dataset.","PeriodicalId":138931,"journal":{"name":"2022 IEEE 12th International Conference on Consumer Electronics (ICCE-Berlin)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125361480","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：