@WWW.SASTRA.EDU
ASSISTANT PROFESSOR III , SCHOOL OF COMPUTING
SASTRA DEEMED UNIVERSITY
ARTIFICIAL INTELLIGENCE, MACHINE LEARNING, DEEP LEARNING AND COMPUTER VISION
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
G. Revathy, S. Senthilvadivu, S. D. Prabu Ragavendiran, and D. Sathya
AIP Publishing
P. Aurchana, R. Indhumathi, G. Revathy, and A. Ramalingam
IGI Global
Emotion recognition refers to the process of identifying the emotions expressed by an individual, typically through their facial expressions, speech, body language, and sometimes physiological signals like heart rate or skin conductance. In this chapter, facial expression is used to recognise. Emotions like happiness, sadness, anger, fear, surprise, and disgust are typically recognized. This chapter aims at developing a real-time approach to classification of facial emotions such as happy, normal, yawn, and sleep in a real-time context. For this, images are captured using sensors and stored in a cloud storage bucket in which the processing is done. The facial emotions are identified through the use of Haar cascade classifiers. The histogram-oriented gradients features are extracted in the detected facial emotion images, and the extracted features are classified by using machine learning models support vector machine and k-nearest neighbour classifiers as happy, normal, yawn, and sleep. The suggested system outperforms other current systems when tested with real-time datasets.
G. Revathy, M. Vengateshwaran, M. Revathi, and K. V. Priyadharshini
AIP Publishing
G. Indirani, G. Revathy, Suresh Kumar Ramu Ganesan, and P. G. Palanimani
IGI Global
Medical professionals who work in the field of heart disease have their own set of limitations, and they can only anticipate heart attacks with a 67% accuracy rate. Doctors require a support system to better forecast heart disease in today's epidemic condition. This chapter describes the architecture for checking heart rate and other data monitoring approaches, as well as how to leverage machine learning techniques. One example of it is random forest classification algorithm to forecast heart attacks using gathered heart rate data and other health-related information. The methodology employed in this chapter is data gathering utilising IoT sensors for post-COVID-19 patients, and the patients' risk of heart attack is forecasted. The chosen random forest algorithm has a 93% accuracy rating.
G. Revathy, J. Jeyabharathi, Madonna Arieth, and A. Ramalingam
IGI Global
Food security is a major problem worldwide. Ensuring that the crops produced are both safe and wholesome is crucial not only for people as the ultimate consumers of the crops, but also for farmers. Plant diseases are responsible for a significant percentage of crop losses. This alleviates the need for a fast and accurate model to discriminate and identify plants with diseases. The base chapter chosen aims to achieve the same through deep learning. The data set used in the work was obtained from Plant Village Dataset. The work customs deuce pre-trained models, EfficientNetB0 and DenseNet121, to citation the traits of the plants. The extracted traits are then fused together through concatenation to allow the model to read the more meaningful crop trait data. This also ensures that the different sets of feature data read by the two models compensate for any feature loss during extraction. It turns out that the above method gives better results associated to other models.
G. Revathy, J. Jeyabharathi, Madonna Arieth, and A. Ramalingam
IGI Global
Food security is a major problem worldwide. Ensuring that the crops produced are both safe and wholesome is crucial not only for people as the ultimate consumers of the crops, but also for farmers. Plant diseases are responsible for a significant percentage of crop losses. This alleviates the need for a fast and accurate model to discriminate and identify plants with diseases. The chapter aims to achieve the same through deep learning. The data set used in the work was obtained from Plant Village Dataset. The work customs deuce pre-trained models, EfficientNetB0 and DenseNet121, to citation the traits of the plants. The extracted traits are then fused together through concatenation to allow the model to read the more meaningful crop trait data. This also ensures that the different sets of feature data read by the two models compensate for any feature loss during extraction. It turns out that the above method gives better results associated to other models.
G. Revathy, E. Gurumoorthi, C. Sasikala, and T. M. Latha
AIP Publishing
C. Sasikala, S. Prabakaran, S. D. Prabu Ragavendiran, V. Gomathi, and G. Revathy
AIP Publishing
S. Chandrakala, K. Deepak, and G. Revathy
Springer Science and Business Media LLC
R. Sathya, S Senthilvadivu, S. Ananthi, V.C. Bharathi, and G Revathy
IEEE
Plant leaf infection recognition using supervised machine learning has emerged as a promising solution to address the pressing challenges in agriculture and plant pathology. This innovative approach leverages supervised learning techniques to develop robust models capable of accurately identifying diseases and abnormalities in plant leaves based on input images. The proposed process involves several key steps. Initially, a diverse real time data's of brinjal images containing both infected and normal plant leaf is collected and meticulously labeled. The real time dataset covered healthy brijal leafs (HL), Cercospora solani(CS) diseases, Tobacco Mosaic Virus (TMV) diseases, Pythium aphanidermatum (PA) diseases, Pseudomonas solanacearum (PS) deseases and Alternaria melongenea (AM) diseases. Data pre-processing stage, such as filtering, noise removal, resizing and extraction are then evaluated to ensure consistency and enhance the dataset's diversity. Next, meaningful information are taken out from the preprocessed brinjal frames to serve as inputs for the machine learning model. Leaf Intensity Vector (LIV) + Principle Component Analysis + Gray Level Co-occurrence Matrix (GLCM) + Support Vector Machine are employed for brinjal leaf disease reorganization. Finally, the extracted proposed features are classified using Polynomial and RBF kernel of SVM, KNN, Random Forests (RF) and Decision Trees (DTs). The performance of the proposed brinjal leaf diseases classification system gives higher accuracy of SVM RBF (98.48%) on brinjalleaf disordered models.
N. S. Kavitha, M. Sakthivel, B Sreedevi, and G Revathy
IEEE
In recent years, pollutants have affected the water quality. As a result, water quality modeling and prediction are becoming increasingly important for minimizing water pollution. The Water Quality Index (WQI) and Water Quality Classification (WQC) are calculated using advanced machine learning (ML) algorithms. Several people are currently suffering from serious diseases caused by contaminated water. The proposed water quality monitoring system utilizes the water quality data to determine the water quality. The main purpose is to predict water quality using a machine learning system. Water resource management is so critical for improving water quality. Water pollution can be properly controlled if data are reviewed and water prominence is projected in development. Numerous existing studies have addressed this topic; nevertheless, additional research on the utility, reliability, accuracy, and usability of current water quality control techniques is required. Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Regression Analysis are used to evaluate the model's performance.
G. Revathy, G. Indirani, K. Senthilvadivu, D Sathya, A. Kalaiyarasi, and A. Ramalingam
IEEE
Human error-related traffic accidents result in an increasing number of fatalities and injuries worldwide each year. It has been determined that driving when fatigued is a major contributor to auto accidents. Growing weariness has been shown to impair driving performance, and the collisions that result from this impairment account for more than 20% of all motor accidents. The goal of the driver drowsiness detection system is to increase road safety by reducing the frequency of accidents caused by sleepy drivers. Accidents caused by tiredness have increased recently. Many facial expressions, such as fatigue in the eyes, can be used to detect drowsiness. The suggested device watches the driver's eyes and uses a webcam pointed directly at the driver's face to identify signs of fatigue. A warning signal is issued to the driver to alert them when drowsiness is detected.
G. Revathy, U. Sesadri, Shaji. Theodore, J. Justina Princy Thilagavathy, S. Senthilvadivu, and V. Senthil Murugan
IEEE
A stroke is the outcome of an abrupt cessation of blood flow to a region of the brain. Liable on the fragment of the brain that has been hurt, disability is caused by a loss of blood flow because brain cells gradually perish. In order to forecast stroke and maintain a healthy lifestyle, early symptom detection might be very beneficial. In order to afford a strong substance for the long-term peril prediction of stroke incidence, a quantity of replicas are developed and evaluated using machine learning (ML) in this study. The main contribution of this study is an ensemble, random forest, SVM, and XgBoost method that performs well and is validated by a variety of system of measurement, such as precision, recall, F-measure, and accuracy. According to the results of the experiment, random forest, XgBoost, SVM, and random forest classification have an accuracy of 96%, outperforming the other methods. Last but not least, it is recommended to take a number of preventative steps to lessen the risk of having a stroke, such as quitting smoking and abstaining from alcohol.
G. Revathy, Pokkuluri Kiran Sree, S. Sasikala Devi, R. Karunamoorthi, and S. Senthil Vadivu
Springer Nature Singapore
G. Revathy, S. Venkateswaran, V. Senthil Murugan, V. Devi, A. Mohanadevi, and G. Saravanan
IEEE
Cardiovascular disease is among the conditions that pose the greatest risk to life. Nearly 17 million people die as a result of its high mortality rate worldwide. To treat the illness quickly and reduce mortality, early diagnosis is helpful. The occurrence and absenteeism of the ailment can be hush-hush consuming a variability of ML techniques. The UCI dataset is to categorize heart disease using the techniques of LR, NB,SVM and Convolution Neural Networks(CNN). To progress the model’s recital, the dataset was cleaned, missing value searches were carried out, and feature selection was done through correlation by the goal worth for all of the features. The traits with the highest favourable associations were picked. The dataset is then alienated into train and trial sets, and classification is completed experimenting with a 70:30:80 ratio. The most accurate dividing ratio is 80:20. The best outcome will be recorded and used in the suggested model, which will compare Logistic regression, Naive Bayes, and Support vector machines with and without feature selection. Among all the models CNN shows the best result.
S. Chandrakala and G. Revathy
CRC Press
P.Muruga Priya, Sudhakaran Krishnan, G Revathy, L. Kalaiselvi, and Ms.T. Usha
IEEE
The DNA microarray technology has modernized the approach of biology research in such a way that scientists can now measure the expression levels of thousands of genes simultaneously in a single experiment Gene expression profiles represent the state of a cell at a molecular level have great potential as a medical diagnosis tool. Diseases classification with gene expression data is known to include the keys for addressing the fundamental harms relating to diagnosis and discovery. The recent introduction of DNA microarray technique has complete simultaneous monitoring of the large number of gene expressions possible. With this large quantity of gene expression data, experts have started to discover the possibilities of disease classification using gene expression data. Quite a large number of methods have been planned in recent years with hopeful results. But there are still a set of issues, which needs to be addressed. In order to gain insight into the disease classification difficulty, it is necessary to get a closer look at the problem, the proposed solutions and the associated issues all together. In this project, we present a comprehensive searching method, clustering method and classification method such as Pattern similarity search, Spatial Expectation Maximization, K nearest neighbor classification and estimate them based on their evaluation time, classification accuracy and ability to reveal biologically meaningful gene information. Based on the multiclass classification method, the diagnosis the diseases such as Cancer (Lung, Blood, Breast, and Skin) diseases and other diseases also find severity levels of diseases and also prescribe the medicine for affected diseases. The proposed experimental results show the classifier performance through graphs with improved accuracy.
G Revathy, P.Muruga Priya, K. Senthilnathan, P. Mythili, and S.V. Haridharani
IEEE
Heart disease is a serious health problem that has afflicted a lot of people all over the world. In our work, we have proposed a GUI-based machine learning-based approach that is efficient and accurate for identifying heart illness. An correct diagnosis and the right treatment can save several lives. This testing method not only costs a lot of money, but the results also fail to correctly identify HD patients.
B. Sreedevi, Durga Karthik, J. Glory Thephoral, M. Jeya Pandian, and G. Revathy
Springer Nature Singapore
G. Revathy, K. Selvakumar, P. Murugapriya, and D. Ravikumar
CRC Press
G. Revathy, Saleh A. Alghamdi, Sultan M. Alahmari, Saud R. Yonbawi, Anil Kumar, and Mohd Anul Haq
Elsevier BV
G. Revathy, K. Bhavana Raj, Anil Kumar, Spurthi Adibatti, Priyanka Dahiya, and T.M. Latha
Elsevier BV
Shaji. K. A. Theodore, K. Selvakumar, and G. Revathy
Springer Nature Singapore
G. Revathy, Durga Karthik, and B. Sreedevi
IEEE
In recent years, Internet of Things enabling applications, which have provided excellent answers to a variety of challenges. This fast-growing industry is led by wireless sensor networks. Smart medical devices and wearables, for example, play an important part in the Internet of Things, as they may collect a variety of longitudinal patient-generated health data while also presenting preliminary diagnosis options. As part of their efforts to serve patients with IoT-based solutions, experts apply ml to give effective resolutions in bleeding detection. This work describes a smart IoT-based solution for human brain hemorrhage diagnostics that uses deep learning algorithms to reduce death rates and provide correct treatment recommendations. The SVM and Recurrent Neural Network were used to classify the images from the computed tomography scans for the intracranial dataset. When compared to prior techniques such as naive bayes, KNN, and K-medoids, the classification results for the SVM and Recurrent neural network are high. According to the findings, the recurrent neural network beats other methods for identifying intracranial images. The output of the classification tool offers information on the type of brain hemorrhage, which helps to validate an expert’s diagnosis and is utilized as a learning tool for trainee radiologists to eliminate errors in existing systems.