Lung cancer has become of the major health issues in recent year. Early detection and remedy is very important to reduce the chances of death of the sufferers. In this work, Gabor filter image processing has been used to reduce the noise in the images received from the data set along with watershed segmentation to define the image. Features such as mean, standard deviation and energy are found in the clusters of the Lung CT images. RMS, skewness, etc. are also attained. A trained model is created using the extracted features and fed to a support vector machine. An accuracy of 94% has been achieved in the classification of early lung cancer detection. A variety of image processing techniques has been employed to detect the pulmonary cellular breakdown. This research will assist the medical practitioner to diagnose lung cancer at early stages in future,
{"title":"Graphical User Interface based platform for the Lung Cancer Classification","authors":"Shreyansh Kumar Gautam, Saurabh Pandey, Saurabh Kumar Sinha, Kirti","doi":"10.1109/ESCI53509.2022.9758246","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758246","url":null,"abstract":"Lung cancer has become of the major health issues in recent year. Early detection and remedy is very important to reduce the chances of death of the sufferers. In this work, Gabor filter image processing has been used to reduce the noise in the images received from the data set along with watershed segmentation to define the image. Features such as mean, standard deviation and energy are found in the clusters of the Lung CT images. RMS, skewness, etc. are also attained. A trained model is created using the extracted features and fed to a support vector machine. An accuracy of 94% has been achieved in the classification of early lung cancer detection. A variety of image processing techniques has been employed to detect the pulmonary cellular breakdown. This research will assist the medical practitioner to diagnose lung cancer at early stages in future,","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"409 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121815852","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758373
Vighnesh Pathrikar, Tejas Podutwar, S. Vispute, Akshay Siddannavar, Akash Mandana, K. Rajeswari
On January 30, 2020, the World Health Organisation classified the Covid-19 outbreak a Public Health Emergency of International Concern, and a pandemic was proclaimed on March 11, 2020. Two years after the Covid-19 outbreak, the virus has new transmutations plus is turning out to be more difficult for forecasting in terms of both its behaviour and severity. Various techniques for time series analysis of coronavirus (Covid-19) cases were examined in this study. The Deep Learning model chosen, Long Short-Term Memory (LSTM) is compared against Statistical approaches, such as Linear Regression, Auto-Regressive Integrated Moving Average (ARIMA), and Seasonal Auto-Regressive Integrated Moving Average (SARIMA), based on a variety of performance metrics. Following the estimates of the superior algorithm, medical care professionals can act at the appropriate moment to supply Equipment to health care institutions and further help the public. According to our data, as the number of projected days grows, so does the model's error rate. Forecasted trends also suggest that statistical approaches are relatively better overall for predictions of fewer days, but Deep Learning methods are relatively better for forecasts of more days.
{"title":"Forecasting Diurnal Covid-19 Cases for Top-5 Countries Using Various Time-series Forecasting Algorithms","authors":"Vighnesh Pathrikar, Tejas Podutwar, S. Vispute, Akshay Siddannavar, Akash Mandana, K. Rajeswari","doi":"10.1109/ESCI53509.2022.9758373","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758373","url":null,"abstract":"On January 30, 2020, the World Health Organisation classified the Covid-19 outbreak a Public Health Emergency of International Concern, and a pandemic was proclaimed on March 11, 2020. Two years after the Covid-19 outbreak, the virus has new transmutations plus is turning out to be more difficult for forecasting in terms of both its behaviour and severity. Various techniques for time series analysis of coronavirus (Covid-19) cases were examined in this study. The Deep Learning model chosen, Long Short-Term Memory (LSTM) is compared against Statistical approaches, such as Linear Regression, Auto-Regressive Integrated Moving Average (ARIMA), and Seasonal Auto-Regressive Integrated Moving Average (SARIMA), based on a variety of performance metrics. Following the estimates of the superior algorithm, medical care professionals can act at the appropriate moment to supply Equipment to health care institutions and further help the public. According to our data, as the number of projected days grows, so does the model's error rate. Forecasted trends also suggest that statistical approaches are relatively better overall for predictions of fewer days, but Deep Learning methods are relatively better for forecasts of more days.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124636549","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 : 2022-03-09DOI: 10.1109/esci53509.2022.9758189
H. Secretary, Shri Malojiraje
of
的
{"title":"Message from Vice Chancellor","authors":"H. Secretary, Shri Malojiraje","doi":"10.1109/esci53509.2022.9758189","DOIUrl":"https://doi.org/10.1109/esci53509.2022.9758189","url":null,"abstract":"of","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"75 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124638009","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758299
A. Muduli, Malladi Sri Lalitha Sandhya Gayatri
The paper presents a rectangular patch antenna with an impedance band ranging from 1.5 to 8.7 GHz. The proposed antenna operates for five bands which cover many wireless applications. The antenna is designed on an FR4 substrate with a dielectric constant of 4.4, and multiband operations are achieved by inserting circular slots into the rectangular patch. Different parameters like Directivity, VSWR, Return Loss and gain are studied. The simulation and measured results describe the performance of multiband antenna required for 5G Communication.
{"title":"A Modified Multiband Antenna for 5G Communication","authors":"A. Muduli, Malladi Sri Lalitha Sandhya Gayatri","doi":"10.1109/ESCI53509.2022.9758299","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758299","url":null,"abstract":"The paper presents a rectangular patch antenna with an impedance band ranging from 1.5 to 8.7 GHz. The proposed antenna operates for five bands which cover many wireless applications. The antenna is designed on an FR4 substrate with a dielectric constant of 4.4, and multiband operations are achieved by inserting circular slots into the rectangular patch. Different parameters like Directivity, VSWR, Return Loss and gain are studied. The simulation and measured results describe the performance of multiband antenna required for 5G Communication.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114540794","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758354
H. K. Bhuyan, Biswajit Brahma, S. Nyamathulla, S. Mohapatra
Feature selection has been emphasized on an operative approach for dealing with large volume data. The majority of these approaches are skewed into high-ranking features to get well right features towards classification. This paper proposes a structured feature ranking (SFR) approach for large volume data to address this challenge. We present a subspace feature-based clustering approach to find out feature-based cluster as per class labels. The various feature clusters are created ranked for features independently using the SFR approach, based on the subspace weight provided by SFC. Then, for ranking the features, we offer a structured feature weighting method in which the high-rank characteristics are utilized for class labels. SFC's approach has been tested in a variety of features. On a collection of large volume datasets, the proposed SFR approach is compared to six feature selection methods. The results demonstrate that SFR method outperformed than methods.
{"title":"Structured Ranking Method-based Feature Selection in Data Mining","authors":"H. K. Bhuyan, Biswajit Brahma, S. Nyamathulla, S. Mohapatra","doi":"10.1109/ESCI53509.2022.9758354","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758354","url":null,"abstract":"Feature selection has been emphasized on an operative approach for dealing with large volume data. The majority of these approaches are skewed into high-ranking features to get well right features towards classification. This paper proposes a structured feature ranking (SFR) approach for large volume data to address this challenge. We present a subspace feature-based clustering approach to find out feature-based cluster as per class labels. The various feature clusters are created ranked for features independently using the SFR approach, based on the subspace weight provided by SFC. Then, for ranking the features, we offer a structured feature weighting method in which the high-rank characteristics are utilized for class labels. SFC's approach has been tested in a variety of features. On a collection of large volume datasets, the proposed SFR approach is compared to six feature selection methods. The results demonstrate that SFR method outperformed than methods.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"27 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127979846","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758178
Rohan S Benhal, T. Parbat, Honey Jain
Smart Grids are electricity networks with two-way power & data flow capabilities. This allows them to measure, actuate & repair grid anomalies arising due to usage variation, short-circuits, and other issues. These grids work using multiple small power producers that utilize solar, wind, and biogas, along with other conventional sources of energy. Due to which these grids are decentralized in nature, and include small-scale transmission & regional supply compensation. Thus, these grids work in both directions (from supply to consumer, and consumer to supply), which is facilitated by active participation of consumers. In order to manage such a complex infrastructure, a wide variety of smart grid deployment models are proposed by researchers over the years. These models vary in terms of grid size, capacity, deployment cost, power efficiency, area of application, etc. Furthermore, these models also vary largely in terms of performance, usability features, and internal working operations. Due to such a wide variation, it is difficult for researchers and grid designers to select the most optimum model(s) for their deployments. In order to reduce the complexity of model selection, this text reviews some of the most recently proposed smart grid deployment models, and discusses their advantages, nuances, limitations and future research scopes. This text majorly focusses smart grid design from a demand side perspective, and also compares the reviewed models in terms of statistical parameters including complexity of deployment, cost of deployment, and power efficiency. This statistical comparison will assist readers to select the most optimum model(s) for context specific use. Moreover, this text also recommends various fusion mechanisms which can be utilized by researchers & grid designers to combine internal working architectures of reviewed models. These fusion models are capable of combining best design practices observed from the reviewed models, and assist in further improving smart grid deployments.
{"title":"An Empirical Analysis of Smart Grid Deployment System Models Based on Demand Side Perspective","authors":"Rohan S Benhal, T. Parbat, Honey Jain","doi":"10.1109/ESCI53509.2022.9758178","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758178","url":null,"abstract":"Smart Grids are electricity networks with two-way power & data flow capabilities. This allows them to measure, actuate & repair grid anomalies arising due to usage variation, short-circuits, and other issues. These grids work using multiple small power producers that utilize solar, wind, and biogas, along with other conventional sources of energy. Due to which these grids are decentralized in nature, and include small-scale transmission & regional supply compensation. Thus, these grids work in both directions (from supply to consumer, and consumer to supply), which is facilitated by active participation of consumers. In order to manage such a complex infrastructure, a wide variety of smart grid deployment models are proposed by researchers over the years. These models vary in terms of grid size, capacity, deployment cost, power efficiency, area of application, etc. Furthermore, these models also vary largely in terms of performance, usability features, and internal working operations. Due to such a wide variation, it is difficult for researchers and grid designers to select the most optimum model(s) for their deployments. In order to reduce the complexity of model selection, this text reviews some of the most recently proposed smart grid deployment models, and discusses their advantages, nuances, limitations and future research scopes. This text majorly focusses smart grid design from a demand side perspective, and also compares the reviewed models in terms of statistical parameters including complexity of deployment, cost of deployment, and power efficiency. This statistical comparison will assist readers to select the most optimum model(s) for context specific use. Moreover, this text also recommends various fusion mechanisms which can be utilized by researchers & grid designers to combine internal working architectures of reviewed models. These fusion models are capable of combining best design practices observed from the reviewed models, and assist in further improving smart grid deployments.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"92 7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131248501","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758329
Prerana P. Autade, S. Turkane, A. Deshpande
The power dissipation in the electronic products needs to be lowered to conserve the battery life and reliable operations. To reduce power dissipation in various levels such as algorithmic level, architectural level and circuit level, the researchers have been concentrating. To stay away from energy dispersal in a circuit, it is planned utilizing reversible processing. Reversible figuring is an interaction where the info data can be created back from its yield data. The early explores have been focused on the actual reversibility, the main kind of reversibility. Actual reversibility is an interaction which should result in no expansion in actual entropy. To accomplish this, an actual machine is required which burns-through zero energy while registering. To fulfil this imperative, the actual machine ought to be non-dissipative and ought to preserve the actual entropy. Consequently the early explores presumed that no actual gadgets can be reversible and theoretical rationale tasks ought to be reversible. Thus it specifies second sort of reversibility, sensible reversibility, in which the data entropy should be moderated. The design is synthesized using reversible gates which are optimized for minimum number of Toffoli gates. The proposed designs are compared with the other designs based on the number of Toffoli gates. Based on the comparison, it can be concluded that the design uses a maximum of 72%less Toffoli gates and a minimum of 1% less Toffoli gates than the designs available in the literature.
{"title":"Design of Multipliers using Reversible Logic and Toffoli Gates","authors":"Prerana P. Autade, S. Turkane, A. Deshpande","doi":"10.1109/ESCI53509.2022.9758329","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758329","url":null,"abstract":"The power dissipation in the electronic products needs to be lowered to conserve the battery life and reliable operations. To reduce power dissipation in various levels such as algorithmic level, architectural level and circuit level, the researchers have been concentrating. To stay away from energy dispersal in a circuit, it is planned utilizing reversible processing. Reversible figuring is an interaction where the info data can be created back from its yield data. The early explores have been focused on the actual reversibility, the main kind of reversibility. Actual reversibility is an interaction which should result in no expansion in actual entropy. To accomplish this, an actual machine is required which burns-through zero energy while registering. To fulfil this imperative, the actual machine ought to be non-dissipative and ought to preserve the actual entropy. Consequently the early explores presumed that no actual gadgets can be reversible and theoretical rationale tasks ought to be reversible. Thus it specifies second sort of reversibility, sensible reversibility, in which the data entropy should be moderated. The design is synthesized using reversible gates which are optimized for minimum number of Toffoli gates. The proposed designs are compared with the other designs based on the number of Toffoli gates. Based on the comparison, it can be concluded that the design uses a maximum of 72%less Toffoli gates and a minimum of 1% less Toffoli gates than the designs available in the literature.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"2015 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131108597","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758191
Nilesh P. Sable, Vijay U. Rathod, P. Mahalle, Dipika R. Birari
A MANET is an entirely devoid-of-infrastructure network. This network is made up of nodes that randomly move around. Since MANET has no central supervision, it can be formed anywhere using randomly moving nodes. This network faces numerous security issues as a result of MANET's vulnerable behaviour. There are numerous security threats to MANET that do not have a solution. It is also difficult to detect these issues. Some security threats are extremely serious. These threats have the potential to bring the network to its knees. Researchers are attempting to determine how to respond to these threats. The NID system is an important tool for protecting MANETs from vulnerabilities and malicious activities. A slew of new techniques have recently been demonstrated; however, due to the continuous launch of the various threats that existing systems are unable to detect, these techniques face significant challenges. The authors have proposed two stage deep learning (TSDL) model in this publication. For efficient NID, a stacked auto-encoder (SAE) with a softmax classifier (SMC) is used. There are two decisive phases in the model: A first phase in the system traffic classification process that uses a possibility score value to determine whether system movement is regular or irregular. This is then used as a bonus feature during the last stage of the decision-making process. Both the normal state and various types of attacks are to be detected, the suggested framework can automatically and efficiently gain knowledge and categories of beneficial feature representations from large amounts of unlabelled data.
{"title":"A Multiple Stage Deep Learning Model for NID in MANETs","authors":"Nilesh P. Sable, Vijay U. Rathod, P. Mahalle, Dipika R. Birari","doi":"10.1109/ESCI53509.2022.9758191","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758191","url":null,"abstract":"A MANET is an entirely devoid-of-infrastructure network. This network is made up of nodes that randomly move around. Since MANET has no central supervision, it can be formed anywhere using randomly moving nodes. This network faces numerous security issues as a result of MANET's vulnerable behaviour. There are numerous security threats to MANET that do not have a solution. It is also difficult to detect these issues. Some security threats are extremely serious. These threats have the potential to bring the network to its knees. Researchers are attempting to determine how to respond to these threats. The NID system is an important tool for protecting MANETs from vulnerabilities and malicious activities. A slew of new techniques have recently been demonstrated; however, due to the continuous launch of the various threats that existing systems are unable to detect, these techniques face significant challenges. The authors have proposed two stage deep learning (TSDL) model in this publication. For efficient NID, a stacked auto-encoder (SAE) with a softmax classifier (SMC) is used. There are two decisive phases in the model: A first phase in the system traffic classification process that uses a possibility score value to determine whether system movement is regular or irregular. This is then used as a bonus feature during the last stage of the decision-making process. Both the normal state and various types of attacks are to be detected, the suggested framework can automatically and efficiently gain knowledge and categories of beneficial feature representations from large amounts of unlabelled data.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"11 6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115038643","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758345
Chaitali Mahajan, Ashish Jadhav
Every year, a lot of people around the world suffer from gun-related violence. A solution for this could be using a Single stage detector to detect such incidents quickly. They provide accurate and fast detection. Normally in single stage detectors YOLOv3tiny provides fast detection than YOLOv3 but with less accuracy. But in this paper when transfer learning is applied to both the versions with the small dataset having new class as gun then tiny version improves with accuracy by 4% than that of v3. When YOLOv3 and tiny version are trained on 3000 and 2500 respectively then we have got that point as a threshold where both gave same accuracy. Their performances were also evaluated using criteria such as precision, recall, F1 score. The key takeaway from this is YOLOv3 tiny performed best in terms of accuracy and F1 score than that of YOLOv3 in case of transfer learning.
{"title":"Gun Detection: Comparative Analysis using Transfer Learning in Single Stage Detectors","authors":"Chaitali Mahajan, Ashish Jadhav","doi":"10.1109/ESCI53509.2022.9758345","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758345","url":null,"abstract":"Every year, a lot of people around the world suffer from gun-related violence. A solution for this could be using a Single stage detector to detect such incidents quickly. They provide accurate and fast detection. Normally in single stage detectors YOLOv3tiny provides fast detection than YOLOv3 but with less accuracy. But in this paper when transfer learning is applied to both the versions with the small dataset having new class as gun then tiny version improves with accuracy by 4% than that of v3. When YOLOv3 and tiny version are trained on 3000 and 2500 respectively then we have got that point as a threshold where both gave same accuracy. Their performances were also evaluated using criteria such as precision, recall, F1 score. The key takeaway from this is YOLOv3 tiny performed best in terms of accuracy and F1 score than that of YOLOv3 in case of transfer learning.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117132569","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 : 2022-03-09DOI: 10.1109/ESCI53509.2022.9758183
V. Kukreja, Vikas Solanki, Anupam Baliyan, Vishal Jain
A transfer learning (TL) based multi-classification model has been developed to classify and recognize collected weather dataset with 1000 different weather images belonging to four different classes of weather. MobileNet V2 has been applied as a pre-trained model with a combination of weather image classifiers which results in the best recognition accuracy of 98.25% in the case of rainy (R) class images. Methodological techniques and challenges encountered while experimenting has also been presented in the detailed description. Along with this, the proposed model has also been compared with a simple convolutional neural network (CNN) model which results in outperformance of the TL model in terms of efficiency and efficacy.
{"title":"WeatherNet: Transfer Learning-based Weather Recognition Model","authors":"V. Kukreja, Vikas Solanki, Anupam Baliyan, Vishal Jain","doi":"10.1109/ESCI53509.2022.9758183","DOIUrl":"https://doi.org/10.1109/ESCI53509.2022.9758183","url":null,"abstract":"A transfer learning (TL) based multi-classification model has been developed to classify and recognize collected weather dataset with 1000 different weather images belonging to four different classes of weather. MobileNet V2 has been applied as a pre-trained model with a combination of weather image classifiers which results in the best recognition accuracy of 98.25% in the case of rainy (R) class images. Methodological techniques and challenges encountered while experimenting has also been presented in the detailed description. Along with this, the proposed model has also been compared with a simple convolutional neural network (CNN) model which results in outperformance of the TL model in terms of efficiency and efficacy.","PeriodicalId":436539,"journal":{"name":"2022 International Conference on Emerging Smart Computing and Informatics (ESCI)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123519477","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：