Pub Date : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101654
I. J. Swarna, Emrana Kabir Hashi
Colon cancer is one of the most prevalent types of cancer. Early diagnosis of colon cancer can lead to an increased chance of successful treatment with less cost. To speed up this process deep learning can provide very useful and effective approaches. In this thesis work, two types of models were developed to classify colon cells from image data - one is the transfer learning model where a deep network Inception V3 is used as the pre-trained model and the other one is an Ensembled model which combines predictions of three simple sequential CNN models. To develop these models, 10k images were used from the LC25000 dataset and a very small Warwick-QU dataset having only 165 images was used to provide new data for retraining and testing purposes. Both models achieved a high result for the first dataset with 99.4% and 99.95% accuracy respectively, where Inception V3 showed 94.545% accuracy on new data from Warwick-QU after retraining and Ensembled model showed 78.182% accuracy. This approach can be used in research in the field of early and effective detection of colon cancer with a larger amount of varying images and more preprocessing methods to reduce overfitting and to make the model perform well in various types of images.
{"title":"Detection of Colon Cancer Using Inception V3 and Ensembled CNN Model","authors":"I. J. Swarna, Emrana Kabir Hashi","doi":"10.1109/ECCE57851.2023.10101654","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101654","url":null,"abstract":"Colon cancer is one of the most prevalent types of cancer. Early diagnosis of colon cancer can lead to an increased chance of successful treatment with less cost. To speed up this process deep learning can provide very useful and effective approaches. In this thesis work, two types of models were developed to classify colon cells from image data - one is the transfer learning model where a deep network Inception V3 is used as the pre-trained model and the other one is an Ensembled model which combines predictions of three simple sequential CNN models. To develop these models, 10k images were used from the LC25000 dataset and a very small Warwick-QU dataset having only 165 images was used to provide new data for retraining and testing purposes. Both models achieved a high result for the first dataset with 99.4% and 99.95% accuracy respectively, where Inception V3 showed 94.545% accuracy on new data from Warwick-QU after retraining and Ensembled model showed 78.182% accuracy. This approach can be used in research in the field of early and effective detection of colon cancer with a larger amount of varying images and more preprocessing methods to reduce overfitting and to make the model perform well in various types of images.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125041084","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101505
Tanvir Raihan Khan, Asif Mostofa, Mrinmoy Dey
Diabetes is a condition that develops when blood glucose, often known as blood sugar, is too high. A diabetic patient must constantly monitor his or her blood glucose level to keep it under control. In the commercial invasive approach, a patient must injure his body part to obtain a blood sample, which is uncomfortable for the patient and can increase the risk of infection. Blood glucose monitoring using a non-invasive technique can lessen discomfort. In this paper, we suggested a non-invasive blood glucose measuring technique that consists of a Near Infrared LED (940nm) and a photodetector to estimate blood glucose levels. In our work, we employed Near Infrared Light to assess blood glucose levels. Following the device's implementation, we compared the accuracies of both diffused reflectance method and diffused transmittance method to see which method is preferable. It was found that diffused transmittance method is the better one of the two. The results from both methods were also compared with a commercial invasive blood glucometer on the market. It is observed from Clarke Error Grid Analysis that, most of the test data from diffused transmittance method lies in Region A. We have also developed an app that can show the data from the devices on patients' smartphones.
{"title":"Non-Invasive Blood Glucose Measurement Device: Performance analysis of Diffused Reflectance method and Diffused Transmittance method using Near Infrared Light","authors":"Tanvir Raihan Khan, Asif Mostofa, Mrinmoy Dey","doi":"10.1109/ECCE57851.2023.10101505","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101505","url":null,"abstract":"Diabetes is a condition that develops when blood glucose, often known as blood sugar, is too high. A diabetic patient must constantly monitor his or her blood glucose level to keep it under control. In the commercial invasive approach, a patient must injure his body part to obtain a blood sample, which is uncomfortable for the patient and can increase the risk of infection. Blood glucose monitoring using a non-invasive technique can lessen discomfort. In this paper, we suggested a non-invasive blood glucose measuring technique that consists of a Near Infrared LED (940nm) and a photodetector to estimate blood glucose levels. In our work, we employed Near Infrared Light to assess blood glucose levels. Following the device's implementation, we compared the accuracies of both diffused reflectance method and diffused transmittance method to see which method is preferable. It was found that diffused transmittance method is the better one of the two. The results from both methods were also compared with a commercial invasive blood glucometer on the market. It is observed from Clarke Error Grid Analysis that, most of the test data from diffused transmittance method lies in Region A. We have also developed an app that can show the data from the devices on patients' smartphones.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"75 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123285965","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101574
Md. Jafril Alam, Sakib Zaman, P. C. Shill, Sujoy Kar, Md. Azizul Hakim
Gastrointestinal(GI) tract cancer is a common type of cancer around the world. Cancer patients require radiotherapy as a part of a cancer diagnosis. To provide therapy in the cancer-affected GI tract, it needs to avoid the stomach and bowels because, in this case, the stomach and intestine are not cancer affected. It is ineffective to manually avoid the intestines and stomach and move the X-ray beam toward the cancer cell because it is a time-consuming, labor-intensive mechanism. Besides these issues, a patient feels uncomfortable while repeatedly X-ray beam is set manually. We implemented a deep learning-based automated medical image segmentation method using LeVit-UNet to overcome these issues. LeVit-UNet is a transformer-based architecture built using the Le Vit unit and CNN. The proposed system properly segments images into three classes: stomach, large, and small bowel. Three backbones of LeVit-UNet: Le Vit-128, Le Vit-192, Le Vit-384 were used in our research. Validation loss, dice score, and IOU were generated and recorded to evaluate all models using three backbones. Though Le Vit-UNet-384 performs well, in our research work, LeVit-UNet-192 performed best.
{"title":"Automated Gastrointestinal Tract Image Segmentation Of Cancer Patient Using LeVit-UNet To Automate Radiotherapy","authors":"Md. Jafril Alam, Sakib Zaman, P. C. Shill, Sujoy Kar, Md. Azizul Hakim","doi":"10.1109/ECCE57851.2023.10101574","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101574","url":null,"abstract":"Gastrointestinal(GI) tract cancer is a common type of cancer around the world. Cancer patients require radiotherapy as a part of a cancer diagnosis. To provide therapy in the cancer-affected GI tract, it needs to avoid the stomach and bowels because, in this case, the stomach and intestine are not cancer affected. It is ineffective to manually avoid the intestines and stomach and move the X-ray beam toward the cancer cell because it is a time-consuming, labor-intensive mechanism. Besides these issues, a patient feels uncomfortable while repeatedly X-ray beam is set manually. We implemented a deep learning-based automated medical image segmentation method using LeVit-UNet to overcome these issues. LeVit-UNet is a transformer-based architecture built using the Le Vit unit and CNN. The proposed system properly segments images into three classes: stomach, large, and small bowel. Three backbones of LeVit-UNet: Le Vit-128, Le Vit-192, Le Vit-384 were used in our research. Validation loss, dice score, and IOU were generated and recorded to evaluate all models using three backbones. Though Le Vit-UNet-384 performs well, in our research work, LeVit-UNet-192 performed best.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128951205","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101545
A. Hossain, Abdul Khaleque, N. Shahriar, Md. Sarwar Hosen, K. Shaha, M. Mizan
In this paper, we propose a broadband metamaterial absorber developed on a simple periodic structure of a U-shaped graphene array that could offer polarization-insensitive wideband terahertz absorption with configurable active tuning. Simulation results reveal that when the graphene's electrochemical potential or Fermi energy was adjusted to 0.7 eV, the bandwidth with absorption greater than 90% is approximately 3.03 THz for transverse electric polarization and 4.3 THz for transverse magnetic polarization while the maximum absorption is greater than 80%. In addition, when the graphene relaxation period is extended from 0.1 ps to 0.5 ps, the same structure functions as a five-band metamaterial absorber with a peak absorption of greater than 90%. Furthermore, the claimed metamaterial absorber provides polarization-insensitive characteristics and retains a high capacity for absorbing both polarized terahertz waves whenever the angle of incidence is less than 40°.
{"title":"Polarization-Insensitive Terahertz Tunable Broadband Metamaterial Absorber on U-shaped Graphene Array","authors":"A. Hossain, Abdul Khaleque, N. Shahriar, Md. Sarwar Hosen, K. Shaha, M. Mizan","doi":"10.1109/ECCE57851.2023.10101545","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101545","url":null,"abstract":"In this paper, we propose a broadband metamaterial absorber developed on a simple periodic structure of a U-shaped graphene array that could offer polarization-insensitive wideband terahertz absorption with configurable active tuning. Simulation results reveal that when the graphene's electrochemical potential or Fermi energy was adjusted to 0.7 eV, the bandwidth with absorption greater than 90% is approximately 3.03 THz for transverse electric polarization and 4.3 THz for transverse magnetic polarization while the maximum absorption is greater than 80%. In addition, when the graphene relaxation period is extended from 0.1 ps to 0.5 ps, the same structure functions as a five-band metamaterial absorber with a peak absorption of greater than 90%. Furthermore, the claimed metamaterial absorber provides polarization-insensitive characteristics and retains a high capacity for absorbing both polarized terahertz waves whenever the angle of incidence is less than 40°.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"1994 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125549528","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10100743
Sharnali Saha, P. C. Shill
Protein secondary structure prediction from amino acid sequences is a challenging and complex task as it has become a must in oder to identifying the similarities/dissimilarities between protein structure. The protein secondary structure is used for studying the biological functionality of species in order to develop new drugs. A sustainable number of research has been done for predicting protein structure but yet the performance is not satisfactory. For this reason, it is necessary and time demanding to develop a technique for predicting protein structure that gives the satisfactory performance for large datasets termed as big datasets. In this article, propose a method based on the support vector machine and fuzzy logic in order to predict protein secondary structure without alignment. In this case, generate the optimal hyper plane of support vector machine using the membership values. Moreover, in order to increase the generalization ability a hybrid kernel support vector machine is propose that gives the better results in terms of classification and learning ability. We have tested the proposed method performance on the several benchmark datasets. The simulation results shows that the proposed technique outperforms better than other existing conventional techniques.
{"title":"Protein Structure Prediction in Structural Genomics without Alignment Using Support Vector Machine with Fuzzy Logic","authors":"Sharnali Saha, P. C. Shill","doi":"10.1109/ECCE57851.2023.10100743","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10100743","url":null,"abstract":"Protein secondary structure prediction from amino acid sequences is a challenging and complex task as it has become a must in oder to identifying the similarities/dissimilarities between protein structure. The protein secondary structure is used for studying the biological functionality of species in order to develop new drugs. A sustainable number of research has been done for predicting protein structure but yet the performance is not satisfactory. For this reason, it is necessary and time demanding to develop a technique for predicting protein structure that gives the satisfactory performance for large datasets termed as big datasets. In this article, propose a method based on the support vector machine and fuzzy logic in order to predict protein secondary structure without alignment. In this case, generate the optimal hyper plane of support vector machine using the membership values. Moreover, in order to increase the generalization ability a hybrid kernel support vector machine is propose that gives the better results in terms of classification and learning ability. We have tested the proposed method performance on the several benchmark datasets. The simulation results shows that the proposed technique outperforms better than other existing conventional techniques.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"49 1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117118820","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101544
MD Ashraf Hossain Ifty, Md. Salim Shahed Shajid
Liver segmentation from computed tomography (CT) images has grown significantly in importance in the field of medical image processing in the last few years. It is the first and most crucial step in any computerized technique for the automatic detection of liver disease, liver volume measurement, and 3D liver volume rendering. The diagnosis and treatment of liver cancer depend heavily on the segmentation of the liver from CT images to get liver volumetric data, but manual segmentation is a strenuous and time-consuming process. The procedure can be accelerated, simplified, and made less error-prone by using deep learning methods. Image segmentation based on deep learning techniques has gained widespread acceptance due to its robustness, efficiency, and it's reproducible nature. Therefore, in this paper, using UNet, MONAI (Medical Open Network for Artificial Intelligence) and PyTorch framework, a deep-learning model to segment the liver from publicly available CT scan dataset was developed. The same ideas that underlie this model for segmenting the liver will allow to create models for segmenting other organs or malignancies using CT data. The goal is to develop a liver segmentation model that can quickly and accurately extract the liver from any given CT image with an accuracy that is on par of manual segmentation performed by a skilled radiologist.
近年来，基于计算机断层扫描(CT)图像的肝脏分割在医学图像处理领域的重要性与日俱增。它是肝脏疾病自动检测、肝脏体积测量和三维肝脏体积绘制等任何计算机技术的第一步，也是最关键的一步。肝癌的诊断和治疗在很大程度上依赖于从CT图像中分割肝脏以获得肝脏体积数据，但人工分割是一个费力且耗时的过程。通过使用深度学习方法，这个过程可以加速、简化，并减少出错的可能性。基于深度学习技术的图像分割由于其鲁棒性、高效性和可重复性而得到了广泛的接受。因此，本文利用UNet、MONAI (Medical Open Network for Artificial Intelligence)和PyTorch框架，开发了一个从公开的CT扫描数据集中分割肝脏的深度学习模型。基于肝脏分割模型的相同思想将允许创建使用CT数据分割其他器官或恶性肿瘤的模型。目标是开发一种肝脏分割模型，该模型可以快速准确地从任何给定的CT图像中提取肝脏，其准确性与熟练的放射科医生进行的人工分割相当。
{"title":"Implementation of Liver Segmentation from Computed Tomography (CT) Images Using Deep Learning","authors":"MD Ashraf Hossain Ifty, Md. Salim Shahed Shajid","doi":"10.1109/ECCE57851.2023.10101544","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101544","url":null,"abstract":"Liver segmentation from computed tomography (CT) images has grown significantly in importance in the field of medical image processing in the last few years. It is the first and most crucial step in any computerized technique for the automatic detection of liver disease, liver volume measurement, and 3D liver volume rendering. The diagnosis and treatment of liver cancer depend heavily on the segmentation of the liver from CT images to get liver volumetric data, but manual segmentation is a strenuous and time-consuming process. The procedure can be accelerated, simplified, and made less error-prone by using deep learning methods. Image segmentation based on deep learning techniques has gained widespread acceptance due to its robustness, efficiency, and it's reproducible nature. Therefore, in this paper, using UNet, MONAI (Medical Open Network for Artificial Intelligence) and PyTorch framework, a deep-learning model to segment the liver from publicly available CT scan dataset was developed. The same ideas that underlie this model for segmenting the liver will allow to create models for segmenting other organs or malignancies using CT data. The goal is to develop a liver segmentation model that can quickly and accurately extract the liver from any given CT image with an accuracy that is on par of manual segmentation performed by a skilled radiologist.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121657780","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101501
Camelia Sinthia, M. H. Kabir
An effective license plate identification algorithm reduces administration expenses while simultaneously enhancing traffic management effectiveness. The novel method suggested in this paper is based on the YOLOv6 amplified convolution model and has two components: Nameplate recognition and location. As a result, the model's receptive field and feature expression capability are improved. For license plate location, CIOU loss takes into account the center distance, aspect ratio, and not just the coverage area of the bounding box. According to the studies, the YOLOv6 model has a 94.7% precision rate for locating license plates, which is 5.6%, 5.1%, and 4.3% better than Faster-RCNN, MobileNet, and the corresponding accuracy rates. We proposed a BLPNET(VGG-19-RESNET-50) model to recognize the characters of number plates and achieved a 100% F1 score.
{"title":"Detection and Recognition of Bangladeshi Vehicles' Nameplates Using YOLOV6 and BLPNET","authors":"Camelia Sinthia, M. H. Kabir","doi":"10.1109/ECCE57851.2023.10101501","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101501","url":null,"abstract":"An effective license plate identification algorithm reduces administration expenses while simultaneously enhancing traffic management effectiveness. The novel method suggested in this paper is based on the YOLOv6 amplified convolution model and has two components: Nameplate recognition and location. As a result, the model's receptive field and feature expression capability are improved. For license plate location, CIOU loss takes into account the center distance, aspect ratio, and not just the coverage area of the bounding box. According to the studies, the YOLOv6 model has a 94.7% precision rate for locating license plates, which is 5.6%, 5.1%, and 4.3% better than Faster-RCNN, MobileNet, and the corresponding accuracy rates. We proposed a BLPNET(VGG-19-RESNET-50) model to recognize the characters of number plates and achieved a 100% F1 score.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124085051","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101661
Amit Roy, Riasat Abdullah, Fahim Ahmed, Shahriar Mashfi, Sazid Hayat Khan, Dewan Ziaul Karim
Retina turns light into pictures and sends messages to the brain. A retinal disease might lead to vision loss or blindness due to eye illness, ocular trauma, or other disorders. Diabetic retinopathy, AMD, and retinal detachment are some widely known retinal-based illnesses. Having eye checkup once a year can assist in preserving the health of the retina. In this matter, the utilization of machine learning and computer vision can be of significant importance. This work proposes an inexpensive, quick approach to diagnose retinal diseases correctly. In today's environment, many people utilize cellphones and high-resolution cameras and hence using computer vision to detect retinal problems will help a lot. This work proposes a lightweight custom CNN model (RetNet) to accurately diagnose and classify retinal disorders. For extensive image recognition, the convolutional neural network was fed 30904 retinal images split into 3 categories: Test, train, and validation. Four retinal conditions: CNV, DME, DRUSEN and NORMAL were deteced and classified. The CNN model trained with these datasets achieved 97.85% training accuracy and 95.41% validation accuracy. Pre-trained models such as Resnet50, InceptionV3, EfficientNetB0, Xception, and VGG16 were also used and their accuracies were 79.34%, 91.32%, 28.0%, 87.94%, and 94.01% respectively. Based on the overall research, it was clear that our lightweight custom CNN model outperformed all pretrained models and produced superior accuracy than the previous works for the used dataset.
{"title":"RetNet: Retinal Disease Detection using Convolutional Neural Network","authors":"Amit Roy, Riasat Abdullah, Fahim Ahmed, Shahriar Mashfi, Sazid Hayat Khan, Dewan Ziaul Karim","doi":"10.1109/ECCE57851.2023.10101661","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101661","url":null,"abstract":"Retina turns light into pictures and sends messages to the brain. A retinal disease might lead to vision loss or blindness due to eye illness, ocular trauma, or other disorders. Diabetic retinopathy, AMD, and retinal detachment are some widely known retinal-based illnesses. Having eye checkup once a year can assist in preserving the health of the retina. In this matter, the utilization of machine learning and computer vision can be of significant importance. This work proposes an inexpensive, quick approach to diagnose retinal diseases correctly. In today's environment, many people utilize cellphones and high-resolution cameras and hence using computer vision to detect retinal problems will help a lot. This work proposes a lightweight custom CNN model (RetNet) to accurately diagnose and classify retinal disorders. For extensive image recognition, the convolutional neural network was fed 30904 retinal images split into 3 categories: Test, train, and validation. Four retinal conditions: CNV, DME, DRUSEN and NORMAL were deteced and classified. The CNN model trained with these datasets achieved 97.85% training accuracy and 95.41% validation accuracy. Pre-trained models such as Resnet50, InceptionV3, EfficientNetB0, Xception, and VGG16 were also used and their accuracies were 79.34%, 91.32%, 28.0%, 87.94%, and 94.01% respectively. Based on the overall research, it was clear that our lightweight custom CNN model outperformed all pretrained models and produced superior accuracy than the previous works for the used dataset.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126196734","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101585
A. Islam, Imranul Khair, Sakawat Hossain, Rashedul Arefin Ifty, M. Arefin, M. Patwary
The importance of agricultural earnings and employment in most countries has decreased with time. That is also true for Bangladesh. Farmers usually design the cultivation process based on their previous experience. Due to a lack of precise agricultural knowledge, they probably end up farming undesirable crops. Several research has employed machine learning methods to forecast agricultural output, but only a few used ensemble machine learning approaches. We use three major crop data which are Aus rice, Aman rice and Potato from the Bangladesh Bureau of Statistics and the seven weather parametrized data from the Bangladesh Meteorological Department over 43 years. The main contribution of this research is the development of an Ensemble Machine Learning Approach (EMLA) by using Catboost Regressor and XGBoost Regressor with their novel combination of Machine Learning Algorithms on the collected dataset. The study compares the accuracy and error rate of the proposed EMLA with eight well-known machine learning algorithms. Our proposed EMLA achieved a high degree of accuracy with R-squared scores of 88.084%,91.776% and 90% respectively for Aus rice, Aman rice and Potato. The results show that the EMLA technique improves the output and prediction by relying on the strong performance of another model. The primary goal of this research is to improve the predictability for overcoming food difficulties and create an intelligent information prediction analysis on farming in Bangladesh for efficient and profitable farming decisions. In this research, we proposed our Ensemble Machine Learning Approach for agricultural crop selection and yield prediction.
{"title":"Ensemble Machine Learning Approach For Agricultural Crop Selection","authors":"A. Islam, Imranul Khair, Sakawat Hossain, Rashedul Arefin Ifty, M. Arefin, M. Patwary","doi":"10.1109/ECCE57851.2023.10101585","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101585","url":null,"abstract":"The importance of agricultural earnings and employment in most countries has decreased with time. That is also true for Bangladesh. Farmers usually design the cultivation process based on their previous experience. Due to a lack of precise agricultural knowledge, they probably end up farming undesirable crops. Several research has employed machine learning methods to forecast agricultural output, but only a few used ensemble machine learning approaches. We use three major crop data which are Aus rice, Aman rice and Potato from the Bangladesh Bureau of Statistics and the seven weather parametrized data from the Bangladesh Meteorological Department over 43 years. The main contribution of this research is the development of an Ensemble Machine Learning Approach (EMLA) by using Catboost Regressor and XGBoost Regressor with their novel combination of Machine Learning Algorithms on the collected dataset. The study compares the accuracy and error rate of the proposed EMLA with eight well-known machine learning algorithms. Our proposed EMLA achieved a high degree of accuracy with R-squared scores of 88.084%,91.776% and 90% respectively for Aus rice, Aman rice and Potato. The results show that the EMLA technique improves the output and prediction by relying on the strong performance of another model. The primary goal of this research is to improve the predictability for overcoming food difficulties and create an intelligent information prediction analysis on farming in Bangladesh for efficient and profitable farming decisions. In this research, we proposed our Ensemble Machine Learning Approach for agricultural crop selection and yield prediction.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127390517","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 : 2023-02-23DOI: 10.1109/ECCE57851.2023.10101673
Anik Das, Md. Mahmudur Rahman, M. A. Matin, N. Amin
Quantum Dot Intermediate Band Solar Cells (QDIBSC) can be a potential candidate in the field of solar cell research. It is an emerging solar cell. Our aim is to find a suitable material for this type of solar cells. Ternary materials are proved very convincing in recent research for solar cells because its bandgap can be varied. InGaN has been chosen as p type and n type material to investigate this solar cell and we found very significant results. InGaN is an emerging solar cell material. The cells had been simulated by varying the band gap of the material. Maximum efficiency is found at 1.21eV. Efficiency at this bandgap is 30.38% ($J_{SC}=47.98 text{mA}/text{cm}^{2}, V_{OC}=0.7429mathrm{V}, FF=0.8524$). Thermal stability also has been investigated of the cell. Normalized efficiency of the cell linearly decreases with the increase of operating temperature at the gradient of −0.14%/°C, which indicates better stability of the cell.
{"title":"Bandgap Analysis of InAs/InGaN Quantum Dot Intermediate Band Solar Cell (QDIBSC)","authors":"Anik Das, Md. Mahmudur Rahman, M. A. Matin, N. Amin","doi":"10.1109/ECCE57851.2023.10101673","DOIUrl":"https://doi.org/10.1109/ECCE57851.2023.10101673","url":null,"abstract":"Quantum Dot Intermediate Band Solar Cells (QDIBSC) can be a potential candidate in the field of solar cell research. It is an emerging solar cell. Our aim is to find a suitable material for this type of solar cells. Ternary materials are proved very convincing in recent research for solar cells because its bandgap can be varied. InGaN has been chosen as p type and n type material to investigate this solar cell and we found very significant results. InGaN is an emerging solar cell material. The cells had been simulated by varying the band gap of the material. Maximum efficiency is found at 1.21eV. Efficiency at this bandgap is 30.38% ($J_{SC}=47.98 text{mA}/text{cm}^{2}, V_{OC}=0.7429mathrm{V}, FF=0.8524$). Thermal stability also has been investigated of the cell. Normalized efficiency of the cell linearly decreases with the increase of operating temperature at the gradient of −0.14%/°C, which indicates better stability of the cell.","PeriodicalId":131537,"journal":{"name":"2023 International Conference on Electrical, Computer and Communication Engineering (ECCE)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122333621","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：