Pub Date : 2022-12-01DOI: 10.1109/OCIT56763.2022.00111
Jagadish Nayak, Smitha Bhat Kaje
A major part of image processing involves convolution process. The pattern recognition techniques which are implemented through Convolutional Neural Networks (CNN) also involves two-dimensional (2D) convolution. The 2D convolution process consists of enormous multiplication operation, which need to be implemented in real time. There is a requirement of fast multiplier for the same operation. Vedic multiplier proved to be faster compared to the conventional multiplication operation. A 2D convolution-based pattern recognition system, which makes use of Vedic Multipliers is proposed in this paper. The proposed system is implemented on Field Programmable Gate Arrays (FPGA) with Verilog programming. The results of Vedic multiplier based convolution and pattern recognition are compared with the conventional multiplier such as Booths algorithm multiplier. The parametric comparison is done in terms of Number of Slice LUT's, Number of Slice Registers, speed and frequency of operation. Results show that there is significant improvement in above said parameters for Vedic multiplier-based convolution in pattern recognition system.
{"title":"Fast Image Convolution and Pattern Recognition using Vedic Mathematics on Field Programmable Gate Arrays (FPGAs)","authors":"Jagadish Nayak, Smitha Bhat Kaje","doi":"10.1109/OCIT56763.2022.00111","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00111","url":null,"abstract":"A major part of image processing involves convolution process. The pattern recognition techniques which are implemented through Convolutional Neural Networks (CNN) also involves two-dimensional (2D) convolution. The 2D convolution process consists of enormous multiplication operation, which need to be implemented in real time. There is a requirement of fast multiplier for the same operation. Vedic multiplier proved to be faster compared to the conventional multiplication operation. A 2D convolution-based pattern recognition system, which makes use of Vedic Multipliers is proposed in this paper. The proposed system is implemented on Field Programmable Gate Arrays (FPGA) with Verilog programming. The results of Vedic multiplier based convolution and pattern recognition are compared with the conventional multiplier such as Booths algorithm multiplier. The parametric comparison is done in terms of Number of Slice LUT's, Number of Slice Registers, speed and frequency of operation. Results show that there is significant improvement in above said parameters for Vedic multiplier-based convolution in pattern recognition system.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115143091","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00106
P. Mohanty, Pushpak Das, D. S. Roy
Since urbanization is occurring at an exponential rate today, energy saving is a key factor for the majority of sustainable smart cities. Out of that, the majority of energy usage is directed toward homes, where there is an enormous possibility for energy optimization. As a result, most academics believe that forecasting this household energy using the advent of AI and machine learning techniques will have social benefits. However, predicting energy consumption alone won't help a city optimize its utilization of energy; it's also crucial to comprehend the factors that influence such predictions so that any available countermeasures can be applied and the city can make decisions about energy optimization that are more accountable, trustworthy, and justifiable to all of its stakeholders. There are different categories of explainers that offer the ability to explore a black box model. Each of these explanations has a connection to a certain model feature. Here, dalex, a Python library that implements a type of explanation, is utilized. a model-neutral user interfaces for interactive fairness and interpretability. It can make machine learning models more understandable. This method is used in this case to know the prediction model and discover the factors responsible for household energy consumption together including their relative importance.
{"title":"Predicting daily household energy usages by using Model Agnostic Language for Exploration and Explanation","authors":"P. Mohanty, Pushpak Das, D. S. Roy","doi":"10.1109/OCIT56763.2022.00106","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00106","url":null,"abstract":"Since urbanization is occurring at an exponential rate today, energy saving is a key factor for the majority of sustainable smart cities. Out of that, the majority of energy usage is directed toward homes, where there is an enormous possibility for energy optimization. As a result, most academics believe that forecasting this household energy using the advent of AI and machine learning techniques will have social benefits. However, predicting energy consumption alone won't help a city optimize its utilization of energy; it's also crucial to comprehend the factors that influence such predictions so that any available countermeasures can be applied and the city can make decisions about energy optimization that are more accountable, trustworthy, and justifiable to all of its stakeholders. There are different categories of explainers that offer the ability to explore a black box model. Each of these explanations has a connection to a certain model feature. Here, dalex, a Python library that implements a type of explanation, is utilized. a model-neutral user interfaces for interactive fairness and interpretability. It can make machine learning models more understandable. This method is used in this case to know the prediction model and discover the factors responsible for household energy consumption together including their relative importance.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124858086","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00110
Dikshya Surabhi Patra, Subhransu Padhee
This manuscript discuss the concept of Text-to-Image synthesis using machine learning methods. For machine learning purpose gradient adversarial network is used. Two different gradient adversarial network namely ControlGAN and ControlGAN-Gradient penalty method are used for the above mentioned task. The inclusion of Gradient-penalty in ControlGAN improves the convergence of the model which is evident from the performance matrices of the system. Microsoft COCO dataset is used for simulation and result validation purposes.
{"title":"Comparative Analysis of ControlGAN and ControlGAN-GP Models based Text-to-Image Synthesis","authors":"Dikshya Surabhi Patra, Subhransu Padhee","doi":"10.1109/OCIT56763.2022.00110","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00110","url":null,"abstract":"This manuscript discuss the concept of Text-to-Image synthesis using machine learning methods. For machine learning purpose gradient adversarial network is used. Two different gradient adversarial network namely ControlGAN and ControlGAN-Gradient penalty method are used for the above mentioned task. The inclusion of Gradient-penalty in ControlGAN improves the convergence of the model which is evident from the performance matrices of the system. Microsoft COCO dataset is used for simulation and result validation purposes.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125925880","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00062
Subhidh Agarwal, Prakhar Rajput, A. Jena
Stock forecasting is one of the most crucial paramount financial techniques which leads to the development of effective stock exchange strategies in the financial world. Stock is considered as the equity of which gives any one as the ownership of that particular corporation. Stock became the current trend for managing the wealth. Stock market plays a major role in economical growth of a developing country. In any country only about 10% of the population engage in stock market. In this work, certain frameworks like ARIMA (Auto Regressive-Integrated-Moving Average), FLANN (Functional Link Artificial Neural Network), ELM (Extreme Learning Machine) models and Grey Wolf optimizer for stock price prediction have been proposed to do the predictions as effectively as possible. The performance of short and long-term predictions of both these models are evaluated with test data and a comparison of minimized errors of both the short and long-term predictions has been presented. The autors have developed a hybrid model using the ELM model and Grey Wolf Optimizer which can be used to change the weights and the number of layers of the ELM model to increase it's accuracy significantly and provide optimum results which are far better when compared to the previous models.
{"title":"A Hybrid Evolutionary model for Stock Price Prediction Using Grey Wolf Optimizer","authors":"Subhidh Agarwal, Prakhar Rajput, A. Jena","doi":"10.1109/OCIT56763.2022.00062","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00062","url":null,"abstract":"Stock forecasting is one of the most crucial paramount financial techniques which leads to the development of effective stock exchange strategies in the financial world. Stock is considered as the equity of which gives any one as the ownership of that particular corporation. Stock became the current trend for managing the wealth. Stock market plays a major role in economical growth of a developing country. In any country only about 10% of the population engage in stock market. In this work, certain frameworks like ARIMA (Auto Regressive-Integrated-Moving Average), FLANN (Functional Link Artificial Neural Network), ELM (Extreme Learning Machine) models and Grey Wolf optimizer for stock price prediction have been proposed to do the predictions as effectively as possible. The performance of short and long-term predictions of both these models are evaluated with test data and a comparison of minimized errors of both the short and long-term predictions has been presented. The autors have developed a hybrid model using the ELM model and Grey Wolf Optimizer which can be used to change the weights and the number of layers of the ELM model to increase it's accuracy significantly and provide optimum results which are far better when compared to the previous models.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"8 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126120209","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00051
Niranjan Panigrahi
Communicating with a person having a hearing or speech disability is always a major challenge. Sign Language (SL) is a medium to remove the barrier of such type of communication. It is a very tough task for a common man to understand SL and interprets its meaning. So, an automated system is necessary which can recognize the SL characters and display its meaning and semantics. In this context, this article has presented a systematic investigation of Artificial Intelligence (AI) based approaches towards examining the difficulties in the classification of characters in Indian Sign Language (ISL). In this work, we adapted ISL recognition using Computer Vision, Machine Learning and Deep Learning methodologies. To achieve this requirement, the captured image undergoes a series of pre-processing steps which include various Computer Vision techniques such as conversion to gray-scale and thresholding using OTSU algorithm. Artificial Neural Network (ANN), Convolutional Neural Network (CNN) and pre-trained models, VGG-19 and Inception-V3using Transfer Learning mechanism are used to train the system. Further, due to large image dataset, the training time of the models are also accelerated using PARAM SHAVAK HPC system which shows a reasonable improvement in the performance of the models.
{"title":"Artificial Intelligence based Indian Sign Language Recognition with Accelerated Performance under HPC Environment","authors":"Niranjan Panigrahi","doi":"10.1109/OCIT56763.2022.00051","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00051","url":null,"abstract":"Communicating with a person having a hearing or speech disability is always a major challenge. Sign Language (SL) is a medium to remove the barrier of such type of communication. It is a very tough task for a common man to understand SL and interprets its meaning. So, an automated system is necessary which can recognize the SL characters and display its meaning and semantics. In this context, this article has presented a systematic investigation of Artificial Intelligence (AI) based approaches towards examining the difficulties in the classification of characters in Indian Sign Language (ISL). In this work, we adapted ISL recognition using Computer Vision, Machine Learning and Deep Learning methodologies. To achieve this requirement, the captured image undergoes a series of pre-processing steps which include various Computer Vision techniques such as conversion to gray-scale and thresholding using OTSU algorithm. Artificial Neural Network (ANN), Convolutional Neural Network (CNN) and pre-trained models, VGG-19 and Inception-V3using Transfer Learning mechanism are used to train the system. Further, due to large image dataset, the training time of the models are also accelerated using PARAM SHAVAK HPC system which shows a reasonable improvement in the performance of the models.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125512023","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00078
Soumyajit Das, Zeeshaan Dayam, P. S. Chatterjee
A classification issue in machine learning is the issue of spotting Distributed Denial of Service (DDos) attacks. A Denial of Service (DoS) assault is essentially a deliberate attack launched from a single source with the implied intent of rendering the target's application unavailable. Attackers typically aims to consume all available network bandwidth in order to accomplish this, which inhibits authorized users from accessing system resources and denies them access. DDoS assaults, in contrast to DoS attacks, include several sources being used by the attacker to launch an attack. At the network, transportation, presentation, and application layers of a 7-layer OSI architecture, DDoS attacks are most frequently observed. With the help of the most well-known standard dataset and multiple regression analysis, we have created a machine learning model in this work that can predict DDoS and bot assaults based on traffic.
{"title":"Application of Random Forest Classifier for Prevention and Detection of Distributed Denial of Service Attacks","authors":"Soumyajit Das, Zeeshaan Dayam, P. S. Chatterjee","doi":"10.1109/OCIT56763.2022.00078","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00078","url":null,"abstract":"A classification issue in machine learning is the issue of spotting Distributed Denial of Service (DDos) attacks. A Denial of Service (DoS) assault is essentially a deliberate attack launched from a single source with the implied intent of rendering the target's application unavailable. Attackers typically aims to consume all available network bandwidth in order to accomplish this, which inhibits authorized users from accessing system resources and denies them access. DDoS assaults, in contrast to DoS attacks, include several sources being used by the attacker to launch an attack. At the network, transportation, presentation, and application layers of a 7-layer OSI architecture, DDoS attacks are most frequently observed. With the help of the most well-known standard dataset and multiple regression analysis, we have created a machine learning model in this work that can predict DDoS and bot assaults based on traffic.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124187811","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00027
Abhyudaya Batabyal, Vinay Singh, Mahendra Kumar Gourisaria, Himansu Das
Nowadays, chronic insomnia is a critical problem of homo-sapiens. An increase in workload and tension in life led to the development of sleep stress. Sleep stress can damage human beings in a physical, psychological, and social manner. Sickness in the stomach, tension, and frayed nerves while sleeping are the most frequent symptoms of sleep stress. Sleep stress can lead to cardiac infarction, depression, senile psychosis, gastrointestinal problems, diabetes, obesity, and emphysematous. This paper primarily focuses on the classification of sleep stress levels using standard machine learning algorithms like Decision Tree (DT), Logistic Regression (LR), Radial basis function Supported-Vector Classifier (RBF-SVC), K-Nearest Neighbor (KNN), Random Forest (RF), Extreme Gradient Boosting (XGB), Linear Support-Vector Classifier (L-SVC), Naive Bayes (NB), Support-Vector Classifier (SVC), on the scaled dataset using Standard Scaling. LR, KNN, and SVC outperformed all the other machine learning classifiers in terms of performance metrics.
如今，慢性失眠是人类的一个严重问题。工作量的增加和生活中的紧张导致了睡眠压力的发展。睡眠压力会对人的身体、心理和社会造成损害。睡觉时胃部不适、紧张和神经紧张是睡眠压力最常见的症状。睡眠压力会导致心脏病、抑郁症、老年性精神病、胃肠道问题、糖尿病、肥胖和肺气肿。本文主要关注使用标准机器学习算法(如决策树(DT)，逻辑回归(LR)，径向基函数支持向量分类器(RBF-SVC)， k -近邻(KNN)，随机森林(RF)，极端梯度增强(XGB)，线性支持向量分类器(L-SVC)，朴素贝叶斯(NB)，支持向量分类器(SVC))对使用标准缩放的缩放数据集进行睡眠压力水平分类。在性能指标方面，LR、KNN和SVC优于所有其他机器学习分类器。
{"title":"Sleep Stress Level Classification through Machine Learning Algorithms","authors":"Abhyudaya Batabyal, Vinay Singh, Mahendra Kumar Gourisaria, Himansu Das","doi":"10.1109/OCIT56763.2022.00027","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00027","url":null,"abstract":"Nowadays, chronic insomnia is a critical problem of homo-sapiens. An increase in workload and tension in life led to the development of sleep stress. Sleep stress can damage human beings in a physical, psychological, and social manner. Sickness in the stomach, tension, and frayed nerves while sleeping are the most frequent symptoms of sleep stress. Sleep stress can lead to cardiac infarction, depression, senile psychosis, gastrointestinal problems, diabetes, obesity, and emphysematous. This paper primarily focuses on the classification of sleep stress levels using standard machine learning algorithms like Decision Tree (DT), Logistic Regression (LR), Radial basis function Supported-Vector Classifier (RBF-SVC), K-Nearest Neighbor (KNN), Random Forest (RF), Extreme Gradient Boosting (XGB), Linear Support-Vector Classifier (L-SVC), Naive Bayes (NB), Support-Vector Classifier (SVC), on the scaled dataset using Standard Scaling. LR, KNN, and SVC outperformed all the other machine learning classifiers in terms of performance metrics.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124515171","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00090
Avijit Mondal, P. S. Chatterjee
Cloud security is a branch of cyber security that focuses on securing cloud computing systems. It is a method for enterprises to use the internet to access storage and virtual services while saving money on infrastructure, considered as the next-generation architecture since it combines application software and databases into big data centers. However, difficulties with confidentiality, protection, integrity, and compliance may surface when the information is not kept, examined, or computed locally. A wide range of data security-related topics are covered by cloud-based data security. For securing an organization's data across the network, data encryption is a common and effective security strategy that is a great alternative. The most vulnerable data is financial and payment systems, and medical data, which can expose customers or clients critical information. Several academics have also offered a range of approaches to safeguards to maintain data privacy and security while it is being transmitted, but there are still a lot of obstacles to the safety of data that is being stored in the cloud. In this paper, several research papers that have been authored and published in this topic are thoroughly examined and analysed. This study gives a review of the literature on different methods for establishing data security in cloud environment.
{"title":"A SYSTEMATIC LITERATURE SURVEY ON DATA SECURITY TECHNIQUES IN A CLOUD ENVIRONMENT","authors":"Avijit Mondal, P. S. Chatterjee","doi":"10.1109/OCIT56763.2022.00090","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00090","url":null,"abstract":"Cloud security is a branch of cyber security that focuses on securing cloud computing systems. It is a method for enterprises to use the internet to access storage and virtual services while saving money on infrastructure, considered as the next-generation architecture since it combines application software and databases into big data centers. However, difficulties with confidentiality, protection, integrity, and compliance may surface when the information is not kept, examined, or computed locally. A wide range of data security-related topics are covered by cloud-based data security. For securing an organization's data across the network, data encryption is a common and effective security strategy that is a great alternative. The most vulnerable data is financial and payment systems, and medical data, which can expose customers or clients critical information. Several academics have also offered a range of approaches to safeguards to maintain data privacy and security while it is being transmitted, but there are still a lot of obstacles to the safety of data that is being stored in the cloud. In this paper, several research papers that have been authored and published in this topic are thoroughly examined and analysed. This study gives a review of the literature on different methods for establishing data security in cloud environment.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124041344","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00048
Aparajita Das, Navajit Saikia, Subhash Ch. Rajbongshi, K. K. Sarma
Human activity recognition is one of the prime focus areas of computer vision having a range of current and evolving applications in the real-world environment such as abnormal activity recognition, pedestrian traffic with action detection, video indexing, gesture recognition, etc. The goal of this paper is to propose a human action recognition framework that can efficiently work in complex background by exploiting the stacked autoencoder principle. Due to the rapid development of artificial intelligence (AI) aided approaches of decision making, deep learning (DL) is a preferred area of research. Among several known DL approaches, the stacked autoencoder has received extensive research interest and is considered to be among the current state-of-the-art approaches. In particular as part of this work, a stacked autoencoder with three hidden layers is trained in the first stage for representation learning. In the second stage, a SoftMax layer is integrated as a final output layer for the classification of various human actions. We applied the proposed method to a publicly available human action database to evaluate its performance. The feasibility and the effectiveness of the proposed stacked autoencoder-based human action recognition framework have been demonstrated by experimental simulation in this paper.
{"title":"Human Activity Recognition based on Stacked Autoencoder with Complex Background Conditions","authors":"Aparajita Das, Navajit Saikia, Subhash Ch. Rajbongshi, K. K. Sarma","doi":"10.1109/OCIT56763.2022.00048","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00048","url":null,"abstract":"Human activity recognition is one of the prime focus areas of computer vision having a range of current and evolving applications in the real-world environment such as abnormal activity recognition, pedestrian traffic with action detection, video indexing, gesture recognition, etc. The goal of this paper is to propose a human action recognition framework that can efficiently work in complex background by exploiting the stacked autoencoder principle. Due to the rapid development of artificial intelligence (AI) aided approaches of decision making, deep learning (DL) is a preferred area of research. Among several known DL approaches, the stacked autoencoder has received extensive research interest and is considered to be among the current state-of-the-art approaches. In particular as part of this work, a stacked autoencoder with three hidden layers is trained in the first stage for representation learning. In the second stage, a SoftMax layer is integrated as a final output layer for the classification of various human actions. We applied the proposed method to a publicly available human action database to evaluate its performance. The feasibility and the effectiveness of the proposed stacked autoencoder-based human action recognition framework have been demonstrated by experimental simulation in this paper.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"124 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130864303","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 : 2022-12-01DOI: 10.1109/OCIT56763.2022.00073
Sabita Sahu, J. Amudha
In many countries, agriculture is the predominant root of income.Agriculture provides food, as well as income to farmers. Maize is one of world's leading crops and universally cultivated as cereal grain. Usually, agricultural specialists or farmers use their skills to identify pests and diseases that affect fruit and leaves on the spot. Even the most experienced farmer is prone to making errors in disease identification while growing crops in a greater scale. To treat leaf disease, pesticides are used, however, this is damaging to people's health [1]. Several Machine learning, Deep learning algorithms are suggested to classify diseases in the maize plant. Identification of maize leaf disease is a great challenge due to environmental changes and illumination variation in weather conditions. This research focuses on using different Deep Learning architectures like optimized DenseNet121,CNN, ResNet50, MobileNet, VGG16, and Inception-V3for classification of maize leaves disease so that preventive measures can be taken by the farmers at early stage to protect the crops. Our proposed optimized Densenet121 model outperformed compared to optimized CNN, and ResNet50 with lesser parameters and higher accuracy.
{"title":"Maize Plant Disease classification using optimized DenseNet121","authors":"Sabita Sahu, J. Amudha","doi":"10.1109/OCIT56763.2022.00073","DOIUrl":"https://doi.org/10.1109/OCIT56763.2022.00073","url":null,"abstract":"In many countries, agriculture is the predominant root of income.Agriculture provides food, as well as income to farmers. Maize is one of world's leading crops and universally cultivated as cereal grain. Usually, agricultural specialists or farmers use their skills to identify pests and diseases that affect fruit and leaves on the spot. Even the most experienced farmer is prone to making errors in disease identification while growing crops in a greater scale. To treat leaf disease, pesticides are used, however, this is damaging to people's health [1]. Several Machine learning, Deep learning algorithms are suggested to classify diseases in the maize plant. Identification of maize leaf disease is a great challenge due to environmental changes and illumination variation in weather conditions. This research focuses on using different Deep Learning architectures like optimized DenseNet121,CNN, ResNet50, MobileNet, VGG16, and Inception-V3for classification of maize leaves disease so that preventive measures can be taken by the farmers at early stage to protect the crops. Our proposed optimized Densenet121 model outperformed compared to optimized CNN, and ResNet50 with lesser parameters and higher accuracy.","PeriodicalId":425541,"journal":{"name":"2022 OITS International Conference on Information Technology (OCIT)","volume":"197 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132741474","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：