Pub Date : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068601
S. Aydin, H. Ş. Bilge
Convolutional neural networks (CNN) show promising results in many fields, especially in computer vision tasks. However, implementing these networks requires computationally intensive operations. Increasing computational workloads makes it difficult to use CNN models in real-time applications. To overcome these challenges, CNN must be implemented on a dedicated hardware platform such as a field-programmable gate array (FPGA). The parallel processing and reconfigurable features of FPGA hardware make it suitable for real-time applications. Nevertheless, due to limited resources and memory units, various optimizations must be applied prior to implementing processing-intensive structures. Both the resources and the memory units used in hardware applications are affected by the data types and byte lengths used to display data. This study proposes arbitrary, precision fixed-point data types for optimal end-to-end CNN hardware implementation. The network was trained on the Central Processing Unit (CPU) to address the classification problem. The CNN architecture was implemented on a Zynq-7 ZC702 evaluation board with a target device xc7z020clg484-1 platform utilizing high level synthesis (HLS) for the inference stage, based on the calculated weight parameters, and predetermined hyperparameters. The proposed implementation produced the results in 0.00329 s based on hardware implementation. In terms of latency metrics, the hardware-based CNN application produced a response approximately 18.9 times faster than the CPU-based CNN application in the inference phase while retaining the same accuracy. In terms of memory utilization and calculation units, the proposed design uses 52% fewer memory units and 68% fewer calculation units than the baseline design. While the proposed method used fewer resources, the classification success remained at 98.9%.
{"title":"Optimal hardware implementation for end-to-end CNN-based classification","authors":"S. Aydin, H. Ş. Bilge","doi":"10.1109/ICITIIT57246.2023.10068601","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068601","url":null,"abstract":"Convolutional neural networks (CNN) show promising results in many fields, especially in computer vision tasks. However, implementing these networks requires computationally intensive operations. Increasing computational workloads makes it difficult to use CNN models in real-time applications. To overcome these challenges, CNN must be implemented on a dedicated hardware platform such as a field-programmable gate array (FPGA). The parallel processing and reconfigurable features of FPGA hardware make it suitable for real-time applications. Nevertheless, due to limited resources and memory units, various optimizations must be applied prior to implementing processing-intensive structures. Both the resources and the memory units used in hardware applications are affected by the data types and byte lengths used to display data. This study proposes arbitrary, precision fixed-point data types for optimal end-to-end CNN hardware implementation. The network was trained on the Central Processing Unit (CPU) to address the classification problem. The CNN architecture was implemented on a Zynq-7 ZC702 evaluation board with a target device xc7z020clg484-1 platform utilizing high level synthesis (HLS) for the inference stage, based on the calculated weight parameters, and predetermined hyperparameters. The proposed implementation produced the results in 0.00329 s based on hardware implementation. In terms of latency metrics, the hardware-based CNN application produced a response approximately 18.9 times faster than the CPU-based CNN application in the inference phase while retaining the same accuracy. In terms of memory utilization and calculation units, the proposed design uses 52% fewer memory units and 68% fewer calculation units than the baseline design. While the proposed method used fewer resources, the classification success remained at 98.9%.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124062726","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068600
Monilal S, Bhama Devi N
This paper proposes a simple rear view aid system based on panoramic vision that can be used to provide the rear-view aid to the driver of a vehicle. The image alignment technique proposed in this paper is very simple, computationally feasible, and is easily extendable for multiple input video streams.
{"title":"A Rear-view Aid System For Vehicles Based on Panoramic Vision","authors":"Monilal S, Bhama Devi N","doi":"10.1109/ICITIIT57246.2023.10068600","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068600","url":null,"abstract":"This paper proposes a simple rear view aid system based on panoramic vision that can be used to provide the rear-view aid to the driver of a vehicle. The image alignment technique proposed in this paper is very simple, computationally feasible, and is easily extendable for multiple input video streams.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126398307","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068688
Ferdianto, Lius Steven Sanjaya, Titan, Erwin Ardianto Halim, Dyah Wahyu Sukmaningsih, A. Effendi, Yulius Lie
Information technology has become the basis for various companies to support operations and management, so companies in various fields have begun to develop information technology within their companies to compete. Based on that phenomenon, there is an opportunity for SMEs also to do the same. Developing information technology within SMEs can solve many problems, such as increasing producvity and managing sales data, inventory, and reports, especially when SMEs have more than one store. This research aims to design applications for Small and Medium Enterprises to support Sales Processes. System Development Life Cycle Methodology was used to develop the application. The requirement-gathering techniques used in this research were interviews, observations with many SMEs, and study literature. The benefit of this research is to help SME owners view and monitor sales data and help the owner make decisions based on data they own.
{"title":"Sales Application Solution for Small Medium Enterprise","authors":"Ferdianto, Lius Steven Sanjaya, Titan, Erwin Ardianto Halim, Dyah Wahyu Sukmaningsih, A. Effendi, Yulius Lie","doi":"10.1109/ICITIIT57246.2023.10068688","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068688","url":null,"abstract":"Information technology has become the basis for various companies to support operations and management, so companies in various fields have begun to develop information technology within their companies to compete. Based on that phenomenon, there is an opportunity for SMEs also to do the same. Developing information technology within SMEs can solve many problems, such as increasing producvity and managing sales data, inventory, and reports, especially when SMEs have more than one store. This research aims to design applications for Small and Medium Enterprises to support Sales Processes. System Development Life Cycle Methodology was used to develop the application. The requirement-gathering techniques used in this research were interviews, observations with many SMEs, and study literature. The benefit of this research is to help SME owners view and monitor sales data and help the owner make decisions based on data they own.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125885427","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068670
Sandeep B. Kadam, V. Abhijith, Premlal Ajikumar Sreelekha
As the popularity of Internet of Things (IoT) devices expands in industries and residences, their low processing power and inadequate security make them ideal targets for attackers. Traditional signature-based methods for detecting malware are inefficient against new malware since a small modification in the malware's source code can modify its signature, making it impossible to detect. Understanding the basics of malware behaviour and combatting hackers requires the classification of malware samples. In this study, we examine an image-based classification of malware in which nine malware families were categorised using a convolution neural network (CNN). Using kfold stratified cross-validation, our model attained a promising 89.5% accuracy in training and 82% accuracy in validation.
{"title":"Visual Based Malware Clustering Using Convolution Neural Network","authors":"Sandeep B. Kadam, V. Abhijith, Premlal Ajikumar Sreelekha","doi":"10.1109/ICITIIT57246.2023.10068670","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068670","url":null,"abstract":"As the popularity of Internet of Things (IoT) devices expands in industries and residences, their low processing power and inadequate security make them ideal targets for attackers. Traditional signature-based methods for detecting malware are inefficient against new malware since a small modification in the malware's source code can modify its signature, making it impossible to detect. Understanding the basics of malware behaviour and combatting hackers requires the classification of malware samples. In this study, we examine an image-based classification of malware in which nine malware families were categorised using a convolution neural network (CNN). Using kfold stratified cross-validation, our model attained a promising 89.5% accuracy in training and 82% accuracy in validation.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125096122","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068707
N. Nautiyal, Piyush Agarwal, Sachin Sharma
The typical method of keeping track of medical records uses paper and centralised systems, which introduces a great deal of uncertainty and can lead to inaccurate information, identity theft, a lack of accessibility, and other problems. With the use of blockchain technology, which offers us security, availability, and integrity, this study seeks to address these problems. Using this, the records can be safeguarded and accessible from anywhere, at any time. This blockchain-based solution can protect a patient's life in an emergency since all of the records can be viewed in one location without the inconvenience of running various tests. Re-chain a secure blockchain-based digital medical health record management system was developed to solve the problems with the centralized and traditional system. It is based on the Ethereum network and uses web3.storage which is an Interplanetary File System (IPFS) system to store files in a decentralized manner.
{"title":"Rechain: A Secured Blockchain-Based Digital Medical Health Record Management System","authors":"N. Nautiyal, Piyush Agarwal, Sachin Sharma","doi":"10.1109/ICITIIT57246.2023.10068707","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068707","url":null,"abstract":"The typical method of keeping track of medical records uses paper and centralised systems, which introduces a great deal of uncertainty and can lead to inaccurate information, identity theft, a lack of accessibility, and other problems. With the use of blockchain technology, which offers us security, availability, and integrity, this study seeks to address these problems. Using this, the records can be safeguarded and accessible from anywhere, at any time. This blockchain-based solution can protect a patient's life in an emergency since all of the records can be viewed in one location without the inconvenience of running various tests. Re-chain a secure blockchain-based digital medical health record management system was developed to solve the problems with the centralized and traditional system. It is based on the Ethereum network and uses web3.storage which is an Interplanetary File System (IPFS) system to store files in a decentralized manner.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128939424","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068675
Shruti Jadon, D. Bhat, Shashank R, Yashaswi Tb, Prasad Hb
Identity management solutions are generally designed to facilitate the management of digital identities and operations such as authentication, and have been widely used in real-world applications. In recent years, there have been attempts to introduce blockchain-based identity management solutions, which allow the user to take over control of his/her own identity (i.e. self-sovereign identity). In this paper, we provide a short review of existing blockchain-based identity management methods. Based on the existing methods, we identify potential gaps and opportunities, and proposed an idea to overcome them using mutable blockchain concept[7] [10]. We have used two algorithms-creation of new identity and updation of existing identities on the blockchain. The main contribution of the proposed algorithm is to store the personal data securely in local database and the address of the corresponding data is stored in the blockchain in hashed format. The results for the proposed approach demonstrate that the mining difficulty and mining time varies with the number of transaction in per block.
{"title":"Mutable Blockchain for Identity Management","authors":"Shruti Jadon, D. Bhat, Shashank R, Yashaswi Tb, Prasad Hb","doi":"10.1109/ICITIIT57246.2023.10068675","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068675","url":null,"abstract":"Identity management solutions are generally designed to facilitate the management of digital identities and operations such as authentication, and have been widely used in real-world applications. In recent years, there have been attempts to introduce blockchain-based identity management solutions, which allow the user to take over control of his/her own identity (i.e. self-sovereign identity). In this paper, we provide a short review of existing blockchain-based identity management methods. Based on the existing methods, we identify potential gaps and opportunities, and proposed an idea to overcome them using mutable blockchain concept[7] [10]. We have used two algorithms-creation of new identity and updation of existing identities on the blockchain. The main contribution of the proposed algorithm is to store the personal data securely in local database and the address of the corresponding data is stored in the blockchain in hashed format. The results for the proposed approach demonstrate that the mining difficulty and mining time varies with the number of transaction in per block.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115385641","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068684
T. Tharun, A. Vamshi, R. Eswari
The vast possibilities of the internet age bring an attraction for artists. As a matter of fact, reports of artist revenue have lowered the potential of music platforms. Streams that are difficult to comprehend, cultural structures inside lack reliable data, inadequate for the digital era, aspects of the music business and a drop in the majority of artists' wages. In this paper, A Non-Fungible Tokens (NFT) in Music Industry using Blockchain Technology are proposed to protect the music copyrights and get revenue rights from untrusted holders. In this model, the Musicians may effortlessly approve and maintain their music rights on a public ledger with the help of the blockchain with no middlemen or any other intermediate business companies.
{"title":"NFT Application for Music Industry using Blockchain Smart contracts","authors":"T. Tharun, A. Vamshi, R. Eswari","doi":"10.1109/ICITIIT57246.2023.10068684","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068684","url":null,"abstract":"The vast possibilities of the internet age bring an attraction for artists. As a matter of fact, reports of artist revenue have lowered the potential of music platforms. Streams that are difficult to comprehend, cultural structures inside lack reliable data, inadequate for the digital era, aspects of the music business and a drop in the majority of artists' wages. In this paper, A Non-Fungible Tokens (NFT) in Music Industry using Blockchain Technology are proposed to protect the music copyrights and get revenue rights from untrusted holders. In this model, the Musicians may effortlessly approve and maintain their music rights on a public ledger with the help of the blockchain with no middlemen or any other intermediate business companies.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128441166","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068723
A.M. Abhishek Sai, Gottimukkala Sahil, Boddu Sasi Sai Nadh, Kalla Likhit Sai Eswar, N. S, K. Prakash, A. Mahesh
Connecting with other people online is common practice and huge amounts of data are generated each day by increasing web activity. This type of data collection can be used to recommend friends on social media. We have utilized Map Reduce and Spark to analyze the vast amount of data. A Friend Recommendation system has been implemented using Map Reduce and Spark. Furthermore, we compared both Distributed Computation techniques in order to determine the optimum solution. We found that spark computation is 16 times faster than Hadoop Map-Reduce computation for Friend Recommendation System. Spark proves to be more efficient than map-reduce in terms of time efficiency.
{"title":"Friend Recommendation System Using Map-Reduce and Spark: A Comparison Study","authors":"A.M. Abhishek Sai, Gottimukkala Sahil, Boddu Sasi Sai Nadh, Kalla Likhit Sai Eswar, N. S, K. Prakash, A. Mahesh","doi":"10.1109/ICITIIT57246.2023.10068723","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068723","url":null,"abstract":"Connecting with other people online is common practice and huge amounts of data are generated each day by increasing web activity. This type of data collection can be used to recommend friends on social media. We have utilized Map Reduce and Spark to analyze the vast amount of data. A Friend Recommendation system has been implemented using Map Reduce and Spark. Furthermore, we compared both Distributed Computation techniques in order to determine the optimum solution. We found that spark computation is 16 times faster than Hadoop Map-Reduce computation for Friend Recommendation System. Spark proves to be more efficient than map-reduce in terms of time efficiency.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126539953","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068579
Subash A, A. Cs, V. M
Internet has become a highly influential medium in recent years and this has led to the arrival of threats like DDoS attacks. DDoS attacks are in rising spree, as attackers are very well resourced. IP spoofing makes the situation further worse by concealing the attacker's identity. The IP Traceback mechanism is effective in identifying the origin of the attack. This paper outlines a hybrid traceback technique with packet marking followed by logging. The proposed scheme utilizes a minimized marking field of 16 bits in an IP header compared to existing techniques. The maximum storage requirement will be around 384 KB (based on CAIDA dataset) for logging on all routers. The path reconstruction has zero false positive and false negative rates. Therefore, the proposed scheme eliminates the packet fragmentation problem by reducing the number of digits for marking with a negligible increase in the storage.
{"title":"An Enhanced Hybrid Scheme for IP Traceback","authors":"Subash A, A. Cs, V. M","doi":"10.1109/ICITIIT57246.2023.10068579","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068579","url":null,"abstract":"Internet has become a highly influential medium in recent years and this has led to the arrival of threats like DDoS attacks. DDoS attacks are in rising spree, as attackers are very well resourced. IP spoofing makes the situation further worse by concealing the attacker's identity. The IP Traceback mechanism is effective in identifying the origin of the attack. This paper outlines a hybrid traceback technique with packet marking followed by logging. The proposed scheme utilizes a minimized marking field of 16 bits in an IP header compared to existing techniques. The maximum storage requirement will be around 384 KB (based on CAIDA dataset) for logging on all routers. The path reconstruction has zero false positive and false negative rates. Therefore, the proposed scheme eliminates the packet fragmentation problem by reducing the number of digits for marking with a negligible increase in the storage.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113959967","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 : 2023-02-11DOI: 10.1109/ICITIIT57246.2023.10068625
Gimhan Rodrigo, Dimanthinie De Silva
The COVID-19 epidemic has altered lifestyles all across the globe, causing people to take additional safety precautions and make using a face mask a requirement. Face masks are becoming more popular, making it occasionally challenging for people to recognize other people. Children and the elderly in particular would have trouble identifying their masked guests, which poses a serious hazard because thieves or burglars would take advantage of the situation. In this study, a system was created using IoT and deep learning technologies that works as a unit to offer a contactless solution to the ongoing COVID-19 pandemic while also enabling home owners to keep track of their visitors and receive notifications when someone comes over. The contactless doorbell was created with the help of a Raspberry Pi and a modified ResNet-50 model using ArcFace loss as the feature extractor to efficiently extract visible features from a masked face and support very accurate recognition. Due to the lack of a real masked face dataset with sufficient data, this study used a data augmentation method to add masks to face images from a dataset. The model was able to achieve a recognition accuracy of 98.27% when evaluated using a masked LFW dataset. Furthermore, testing the face recognition model in real-time with limited users, each with and without a mask yielded an accuracy of 100% in unmasked facial recognition and 90% on masked facial recognition.
{"title":"IoT-enabled Contactless Doorbell with Facial Recognition","authors":"Gimhan Rodrigo, Dimanthinie De Silva","doi":"10.1109/ICITIIT57246.2023.10068625","DOIUrl":"https://doi.org/10.1109/ICITIIT57246.2023.10068625","url":null,"abstract":"The COVID-19 epidemic has altered lifestyles all across the globe, causing people to take additional safety precautions and make using a face mask a requirement. Face masks are becoming more popular, making it occasionally challenging for people to recognize other people. Children and the elderly in particular would have trouble identifying their masked guests, which poses a serious hazard because thieves or burglars would take advantage of the situation. In this study, a system was created using IoT and deep learning technologies that works as a unit to offer a contactless solution to the ongoing COVID-19 pandemic while also enabling home owners to keep track of their visitors and receive notifications when someone comes over. The contactless doorbell was created with the help of a Raspberry Pi and a modified ResNet-50 model using ArcFace loss as the feature extractor to efficiently extract visible features from a masked face and support very accurate recognition. Due to the lack of a real masked face dataset with sufficient data, this study used a data augmentation method to add masks to face images from a dataset. The model was able to achieve a recognition accuracy of 98.27% when evaluated using a masked LFW dataset. Furthermore, testing the face recognition model in real-time with limited users, each with and without a mask yielded an accuracy of 100% in unmasked facial recognition and 90% on masked facial recognition.","PeriodicalId":170485,"journal":{"name":"2023 4th International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127579148","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：