Pub Date : 2021-12-01DOI: 10.1109/uemcon53757.2021.9666659
Cesar Martinez Melgoza, Tyler Groom, Henry Lin, Ameya Govalkar, Kayla Lee, Acacia Codding, K. George
In the age of digital communications, radar receivers prove to be essential for applications involving classification such as air traffic control towers, defense systems, and navigation systems. Detecting Emitters within a Radar Environment presents hurdles to the System Designers such as accounting for interference and trying to classify multiple emitters when they are stacked. This paper presents a few-shot machine learning model that utilizes Siamese networks with classification. Given a relatively small dataset, the Siamese network's task is to find the difference between stacked pulses and normal pulse trains, as well as classify the pulse-descriptor words (PDWs), of the signals in the environment. The PDWs will characterize various aspects of the signal with help from a dynamic-thresholding deinterleaving algorithm. The data for this experiment are laboratory generated signals that are transmitted and received using MATLAB, the Zynq Ultrascale+ MPSoC ZCU104 FPGA board, and the AD-FMCOMMS2-EBZ RF module.
{"title":"Environment Classification and Deinterleaving using Siamese Networks and Few-Shot Learning","authors":"Cesar Martinez Melgoza, Tyler Groom, Henry Lin, Ameya Govalkar, Kayla Lee, Acacia Codding, K. George","doi":"10.1109/uemcon53757.2021.9666659","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666659","url":null,"abstract":"In the age of digital communications, radar receivers prove to be essential for applications involving classification such as air traffic control towers, defense systems, and navigation systems. Detecting Emitters within a Radar Environment presents hurdles to the System Designers such as accounting for interference and trying to classify multiple emitters when they are stacked. This paper presents a few-shot machine learning model that utilizes Siamese networks with classification. Given a relatively small dataset, the Siamese network's task is to find the difference between stacked pulses and normal pulse trains, as well as classify the pulse-descriptor words (PDWs), of the signals in the environment. The PDWs will characterize various aspects of the signal with help from a dynamic-thresholding deinterleaving algorithm. The data for this experiment are laboratory generated signals that are transmitted and received using MATLAB, the Zynq Ultrascale+ MPSoC ZCU104 FPGA board, and the AD-FMCOMMS2-EBZ RF module.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"115 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133046803","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 : 2021-12-01DOI: 10.1109/UEMCON53757.2021.9666604
Deyby Huamanchahua, Jorge Sierra-Huertas, Dana Terrazas-Rodas, Alexander Janampa-Espinoza, Jorge Gonzáles, Sofia Huamán-Vizconde
Upper extremity exoskeletons offer an alternative way to support or rehabilitate patients with physical injury, stroke and spinal cord injury (SCI). This research article presents the kinematic analysis of Exo-First Exoskeleton, which is an 4 DoF upper limb exoskeleton, with the aim of assisting or rehabilitating the shoulder and elbow of the human body. This device covers the entire upper limb of a person, from the clavicle to before the wrist. It is capable of executing motions such as internal-external rotation, adduction-abduction or flexion-extension of the shoulder; and flexion-extension of the elbow. The Denavit-Hartenberg (D-H) method was used to obtain the mathematical model that describes the forward and inverse kinematics of the exoskeleton. Furthermore, the exoskeleton end effector trajectories were obtained using the MATLAB software. The results showed that the proposed design for patients with physical disabilities provides a safer Range of Motion (ROM).
{"title":"Kinematic Analysis of an 4 DOF Upper-Limb Exoskeleton","authors":"Deyby Huamanchahua, Jorge Sierra-Huertas, Dana Terrazas-Rodas, Alexander Janampa-Espinoza, Jorge Gonzáles, Sofia Huamán-Vizconde","doi":"10.1109/UEMCON53757.2021.9666604","DOIUrl":"https://doi.org/10.1109/UEMCON53757.2021.9666604","url":null,"abstract":"Upper extremity exoskeletons offer an alternative way to support or rehabilitate patients with physical injury, stroke and spinal cord injury (SCI). This research article presents the kinematic analysis of Exo-First Exoskeleton, which is an 4 DoF upper limb exoskeleton, with the aim of assisting or rehabilitating the shoulder and elbow of the human body. This device covers the entire upper limb of a person, from the clavicle to before the wrist. It is capable of executing motions such as internal-external rotation, adduction-abduction or flexion-extension of the shoulder; and flexion-extension of the elbow. The Denavit-Hartenberg (D-H) method was used to obtain the mathematical model that describes the forward and inverse kinematics of the exoskeleton. Furthermore, the exoskeleton end effector trajectories were obtained using the MATLAB software. The results showed that the proposed design for patients with physical disabilities provides a safer Range of Motion (ROM).","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133810442","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 : 2021-12-01DOI: 10.1109/uemcon53757.2021.9666647
Dakota Barrios, Tyler Groom, K. George
Teaching a language using tactile vocabulary objects is an effective method of teaching for those with who have communication disabilities such as being blind or deaf. The effectiveness of tactile language learning can be greatly complemented by a tactile communication device, which allows students to easily form sentences then quickly and accurately relay them to the teacher. This paper goes over the design and quantitative results of a tactile communication device specifically based around the inclusion of Radio Frequency Identification (RFID) modules.
{"title":"Design and Implementation of an RFID Based Tactile Communication Device","authors":"Dakota Barrios, Tyler Groom, K. George","doi":"10.1109/uemcon53757.2021.9666647","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666647","url":null,"abstract":"Teaching a language using tactile vocabulary objects is an effective method of teaching for those with who have communication disabilities such as being blind or deaf. The effectiveness of tactile language learning can be greatly complemented by a tactile communication device, which allows students to easily form sentences then quickly and accurately relay them to the teacher. This paper goes over the design and quantitative results of a tactile communication device specifically based around the inclusion of Radio Frequency Identification (RFID) modules.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114360488","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 : 2021-12-01DOI: 10.1109/UEMCON53757.2021.9666483
Md. Abu Obaidah, Faria Soroni, Mohammad Monirujjaman Khan
This paper presents the design and the implementation of an onlinebased babysitting system. This is a web-based babysitting service and information storage system created specifically for urban working families. Since the rate of working women in the country is increasing; Parents are in desperate need of help when it comes to taking care of kids or homeschooling them. This system is designed in an efficient way that connects children or adolescents with parents who need childcare or babysitter services, want to lend a hand. The unique process in our country is capable of providing babysitters as well as there is easy and effective storage of information of all the babysitters and parents who register on the system. The system has a great socio-economic impact on society.
{"title":"Development of an Online Based Babysitting System: Bonne","authors":"Md. Abu Obaidah, Faria Soroni, Mohammad Monirujjaman Khan","doi":"10.1109/UEMCON53757.2021.9666483","DOIUrl":"https://doi.org/10.1109/UEMCON53757.2021.9666483","url":null,"abstract":"This paper presents the design and the implementation of an onlinebased babysitting system. This is a web-based babysitting service and information storage system created specifically for urban working families. Since the rate of working women in the country is increasing; Parents are in desperate need of help when it comes to taking care of kids or homeschooling them. This system is designed in an efficient way that connects children or adolescents with parents who need childcare or babysitter services, want to lend a hand. The unique process in our country is capable of providing babysitters as well as there is easy and effective storage of information of all the babysitters and parents who register on the system. The system has a great socio-economic impact on society.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117152241","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 : 2021-12-01DOI: 10.1109/uemcon53757.2021.9666633
Masoud Malekzadeh, P. Hajibabaee, Maryam Heidari, Samira Zad, Özlem Uzuner, James H. Jones
Text classification is one of the fundamental problems in Natural Language Processing (NLP). Several research studies have used deep learning approaches such as Convolution Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for text classification. Over the past decade, graph-based approaches have been used to solve various NLP tasks including text classification. This paper reviews the most recent state-of-the-art graph-based text classification, datasets, and performance evaluations versus baseline models.
{"title":"Review of Graph Neural Network in Text Classification","authors":"Masoud Malekzadeh, P. Hajibabaee, Maryam Heidari, Samira Zad, Özlem Uzuner, James H. Jones","doi":"10.1109/uemcon53757.2021.9666633","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666633","url":null,"abstract":"Text classification is one of the fundamental problems in Natural Language Processing (NLP). Several research studies have used deep learning approaches such as Convolution Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for text classification. Over the past decade, graph-based approaches have been used to solve various NLP tasks including text classification. This paper reviews the most recent state-of-the-art graph-based text classification, datasets, and performance evaluations versus baseline models.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117298767","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 : 2021-12-01DOI: 10.1109/UEMCON53757.2021.9666709
Robin Chataut, R. Akl, U. K. Dey
Massive MIMO (multiple-input multiple-output) is a sub-6GHz wireless access technology that can provide high spectral and energy efficiency and is considered as one of the key enabling technology for 5G, 6G, and beyond networks. The user signal detection during the uplink is one of the major challenges in massive MIMO systems due to the large number of antennas working together at both the user terminal and the base station. The current iterative methods do not offer great efficiency, and the conventional matrix inversion methods are computationally complex due to the large antennas involved in massive MIMO systems. In this paper, we propose a fast and efficient preconditioned iterative method by introducing a preconditioner based on ICF (Incomplete Cholesky Factorization). Additionally, we introduce a novel matrix initializer to further improve the convergence of the proposed algorithm. The numerical results, when compared to conventional methods, show that the proposed algorithm provides better error performance with optimal computational complexity.
{"title":"An Efficient and Fast-convergent Detector for 5G and Beyond Massive MIMO Systems","authors":"Robin Chataut, R. Akl, U. K. Dey","doi":"10.1109/UEMCON53757.2021.9666709","DOIUrl":"https://doi.org/10.1109/UEMCON53757.2021.9666709","url":null,"abstract":"Massive MIMO (multiple-input multiple-output) is a sub-6GHz wireless access technology that can provide high spectral and energy efficiency and is considered as one of the key enabling technology for 5G, 6G, and beyond networks. The user signal detection during the uplink is one of the major challenges in massive MIMO systems due to the large number of antennas working together at both the user terminal and the base station. The current iterative methods do not offer great efficiency, and the conventional matrix inversion methods are computationally complex due to the large antennas involved in massive MIMO systems. In this paper, we propose a fast and efficient preconditioned iterative method by introducing a preconditioner based on ICF (Incomplete Cholesky Factorization). Additionally, we introduce a novel matrix initializer to further improve the convergence of the proposed algorithm. The numerical results, when compared to conventional methods, show that the proposed algorithm provides better error performance with optimal computational complexity.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124000380","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}
In this paper, a wireless nurse calling system is proposed where any patient can call a nurse for an emergency case, and the notification will be received in the nurse’s wrist band. Currently, most of the emergency calling systems in a hospital are constructed based on hard-wired, which is a costly approach. There have been an attempt to implement the calling system over a wireless network, but the scale of coverage was very tiny. This project is based on a unified WiFi network which highly accessible and cheap to found, thus making it one of the cheapest approaches in this domain. The key component of this project is a WiFi module ESP8266 and a Server. This project can also be used on any kind of scale depending on the needs, ie. auto attendence and location detection. Our proposed system promises to deliver much higher performance and coverage while it is closing the gap between the management and nurses by monitoring calls in real-time.
{"title":"An IOT Based Nurse Calling System for Real-time Emergency Alert Using Local Wireless Network","authors":"Omar Faruk Riyad, Ahraf Sharif, Arif-ur-Rahman Chowdhury Suhan, Mohammad Monirujjaman Khan","doi":"10.1109/uemcon53757.2021.9666705","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666705","url":null,"abstract":"In this paper, a wireless nurse calling system is proposed where any patient can call a nurse for an emergency case, and the notification will be received in the nurse’s wrist band. Currently, most of the emergency calling systems in a hospital are constructed based on hard-wired, which is a costly approach. There have been an attempt to implement the calling system over a wireless network, but the scale of coverage was very tiny. This project is based on a unified WiFi network which highly accessible and cheap to found, thus making it one of the cheapest approaches in this domain. The key component of this project is a WiFi module ESP8266 and a Server. This project can also be used on any kind of scale depending on the needs, ie. auto attendence and location detection. Our proposed system promises to deliver much higher performance and coverage while it is closing the gap between the management and nurses by monitoring calls in real-time.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129695424","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 : 2021-12-01DOI: 10.1109/uemcon53757.2021.9666570
Gan Luan, N. Beaulieu, Xianpeng Wang, Mengxing Huang
A buffer-aided collision resolution scheme for UHF RFID is proposed. The scheme uses buffers to store the collided signal, so collision resolution can be achieved through subtracting the identified signals from the corrupted signal stored in the buffer. Based on the buffer-aided collision resolution technique, a novel buffer-aided dynamic frame-slotted Aloha algorithm with the ability to resolve m-tag-collided slots (B-DFSA-m) is introduced. Simulations show that the system efficiencies of B-DFSA-m with the ability to resolve m = 2, 3, and 4-tag-collided slots are 55%, 64%, 66.5%, and their time efficiencies are 72%, 74%, and 75%. These system and time efficiencies compare favorably with the efficiencies of Q-algorithm, Schoute, MAPP, FEIA, and ILCM, BE-MDT, ds-DFSA, ABTSA, and DBTSA, which are the best previous collision resolution schemes for UHF RFID.
{"title":"Buffer-Aided Collision Resolution for UHF RFID","authors":"Gan Luan, N. Beaulieu, Xianpeng Wang, Mengxing Huang","doi":"10.1109/uemcon53757.2021.9666570","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666570","url":null,"abstract":"A buffer-aided collision resolution scheme for UHF RFID is proposed. The scheme uses buffers to store the collided signal, so collision resolution can be achieved through subtracting the identified signals from the corrupted signal stored in the buffer. Based on the buffer-aided collision resolution technique, a novel buffer-aided dynamic frame-slotted Aloha algorithm with the ability to resolve m-tag-collided slots (B-DFSA-m) is introduced. Simulations show that the system efficiencies of B-DFSA-m with the ability to resolve m = 2, 3, and 4-tag-collided slots are 55%, 64%, 66.5%, and their time efficiencies are 72%, 74%, and 75%. These system and time efficiencies compare favorably with the efficiencies of Q-algorithm, Schoute, MAPP, FEIA, and ILCM, BE-MDT, ds-DFSA, ABTSA, and DBTSA, which are the best previous collision resolution schemes for UHF RFID.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129550269","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 : 2021-12-01DOI: 10.1109/uemcon53757.2021.9666576
Mau-Luen Tham, Y. Wong, Ban-Hoe Kwan, Y. Owada, M. Sein, Yoong Choon Chang
Recent advances in deep learning and computer vision have transformed surveillance into an important application for smart disaster monitoring systems. Based on the detected number of victims and activity of disasters, emergency response unit can dispatch manpower more efficiently, which could save more lives. However, most of existing disaster detection methods fall into the class of single-task learning, which can either detect victim or classify disaster. In contrast, this paper proposes a YOLO-based multi-task model which performs the aforementioned tasks simultaneously. This is accomplished by attaching a disaster classification head model to the backbone of a victim detection model. The head model is inherited from the MobileNetv2 architecture, and we precisely select the backbone feature map layer to which the head model is attached. For the victim detection, results reveal that the solution achieves up to 0.6938 and 20.31 in terms of average precision and frame per second, respectively. Whereas for the disaster classification, the algorithm is comparable with most deep learning models that are specifically trained for single task. This shows that our solution is flexible and robust enough to handle both victim detection and disaster classification.
{"title":"Joint Disaster Classification and Victim Detection using Multi-Task Learning","authors":"Mau-Luen Tham, Y. Wong, Ban-Hoe Kwan, Y. Owada, M. Sein, Yoong Choon Chang","doi":"10.1109/uemcon53757.2021.9666576","DOIUrl":"https://doi.org/10.1109/uemcon53757.2021.9666576","url":null,"abstract":"Recent advances in deep learning and computer vision have transformed surveillance into an important application for smart disaster monitoring systems. Based on the detected number of victims and activity of disasters, emergency response unit can dispatch manpower more efficiently, which could save more lives. However, most of existing disaster detection methods fall into the class of single-task learning, which can either detect victim or classify disaster. In contrast, this paper proposes a YOLO-based multi-task model which performs the aforementioned tasks simultaneously. This is accomplished by attaching a disaster classification head model to the backbone of a victim detection model. The head model is inherited from the MobileNetv2 architecture, and we precisely select the backbone feature map layer to which the head model is attached. For the victim detection, results reveal that the solution achieves up to 0.6938 and 20.31 in terms of average precision and frame per second, respectively. Whereas for the disaster classification, the algorithm is comparable with most deep learning models that are specifically trained for single task. This shows that our solution is flexible and robust enough to handle both victim detection and disaster classification.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128259220","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 : 2021-12-01DOI: 10.1109/UEMCON53757.2021.9666484
Gavin Elliott, Kevin Meehan, Jennifer Hyndman
Domestic violence is a prevalent crime in our society, more so with the introduction of COVID19 restrictions. For the victim, it can be a traumatic experience, so much as to not report the crime. Consequently, the ‘Signal for Help’ hand gestures were recently introduced as a discrete method to enable the victim to confidently express their need for help. This research investigates the classification of these hand gestures using a deep learning approach, which has not previously been implemented in this context. A deep learning approach is chosen due to the favourable results obtained in different contexts on hand gesture classification. Due to the unavailability of a dataset containing images of these hand gestures, a ‘Signal for Help’ dataset containing 112 images is generated as part of this study. These images are pre-processed to be of size 50x50 dimensions. Furthermore, a synthetic version of this dataset is also generated from the pre-processed images containing 2,352 images. The aims of this research are to show that using a synthetic ‘Signal for Help’ dataset improves model performance, and using deep learning is effective in ‘Signal for Help’ hand gesture classification. The results in this research show that using a synthetic ‘Signal for Help’ dataset improves model performance and is effective for ‘Signal for Help’ hand gesture classification.
{"title":"Using CNN and Tensorflow to recognise ‘Signal for Help’ Hand Gestures","authors":"Gavin Elliott, Kevin Meehan, Jennifer Hyndman","doi":"10.1109/UEMCON53757.2021.9666484","DOIUrl":"https://doi.org/10.1109/UEMCON53757.2021.9666484","url":null,"abstract":"Domestic violence is a prevalent crime in our society, more so with the introduction of COVID19 restrictions. For the victim, it can be a traumatic experience, so much as to not report the crime. Consequently, the ‘Signal for Help’ hand gestures were recently introduced as a discrete method to enable the victim to confidently express their need for help. This research investigates the classification of these hand gestures using a deep learning approach, which has not previously been implemented in this context. A deep learning approach is chosen due to the favourable results obtained in different contexts on hand gesture classification. Due to the unavailability of a dataset containing images of these hand gestures, a ‘Signal for Help’ dataset containing 112 images is generated as part of this study. These images are pre-processed to be of size 50x50 dimensions. Furthermore, a synthetic version of this dataset is also generated from the pre-processed images containing 2,352 images. The aims of this research are to show that using a synthetic ‘Signal for Help’ dataset improves model performance, and using deep learning is effective in ‘Signal for Help’ hand gesture classification. The results in this research show that using a synthetic ‘Signal for Help’ dataset improves model performance and is effective for ‘Signal for Help’ hand gesture classification.","PeriodicalId":127072,"journal":{"name":"2021 IEEE 12th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)","volume":"475 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124575430","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：