Dr. N. Anusha
@http:
Associate Professor and Computer Science and Engineering
Vidhya Jyothi Institute of Technology
RESEARCH INTERESTS
Image Processing
Remote Sensing & GIS
Machine learning and Artificial Intelligence
Cloud Computing
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
N. Anusha, Darmoju Deekshitha, Ghadiyaram Bhavya and Buchupalli Mohitha
Asian Research Publishing Network
Movie recommendation systems help movie enthusiasts by suggesting movies to watch without the hassle of having to go through the time-consuming process of deciding from a large collection of movie streaming platforms that recommend movies and TV episodes. News organizations that suggest articles to readers, and online stores that suggest products to customers all benefit from these recommendation systems. The algorithms implemented in this research train their models on the MovieLens dataset and provide users with tailored movie recommendations. The study compares different machine learning algorithms, which include a Content-based model, item-item and user-user collaborative filtering (CF), Collaborative filtering with Singular Value Decomposition (SVD), K Nearest Neighbors, and Non-negative Factorization. The algorithms are evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) to measure their accuracy and performance. While the proposed system which is based on a collaborative approach using SVD determines the connection between various users and, depending on their ratings, recommends movies to others with similar tastes, subsequently allowing users to explore more. The proposed approach using collaborative filtering with SVD performs better with a minimal RMSE of 0. 880258 by giving accurate and appropriate recommendations to the user. The model is further evaluated using performance metrics like Precision, Recall, and f1 score. So, CF with the SVD recommendation model is chosen for implementation and is integrated into a web application that allows the platform users to rate and review the available digital content as well as allows them to restrict screen time using a parental control system. The results of the study in this paper are presented in the form of tables, graphs, and statistical analyses, and can be used to guide the development of new and improved recommendation algorithms.
Shubham P Masurkar, Alpashi L Sadawarti, N. Anusha, T. Swapna, and S. L. Jany Shabu
IEEE
Groundwater is the principal source of water in the area of research. The area covers six villages surrounding the Umred coal mines. Fifteen water samples in total were meticulously collected and a comprehensive investigation was undertaken to study the potential influence of mining activities on the quality of groundwater. Wide range of parameters was investigated and the results were precisely compared to the standards set by the BIS (Bureau of Indian Standards) in 2012 and 2015. All the parameters meet the required standards, except for NO3-, It exceeds the acceptable limits in four samples. The findings show that the samples are well-suited for irrigation as per U.S. Salinity Laboratory Diagram. The Piper’s trilinear diagram for the samples shows the dominance of calcium (Ca2+), magnesium(Mg2+) over sodium (Na+), and potassium (K+), with the weak acid (HCO3-) outweighing strong acids (SO42- + Cl-). Majority of samples in this area lie in the Ca-HCO region. The study also evaluated the water quality of this area, by accessing the following key parameters - pH (Potential of Hydrogen), Mg2+ (Magnesium), TDS (Total Dissolved Solids), TH (Total Hardness), Na+ (Sodium), HCO3- (Bicarbonate), SO42- (Sulphates), Ca2+ (Calcium), Cl- (Chlorides), K+ (Potassium), EC (Electrical Conductivity),NO3- (Nitrates), and F- (Fluorides).
N. Anusha and T. Swapna
IEEE
The proliferation of drones and unmanned aerial vehicles (UAVs) across various sectors, including civil, military, and business applications, has underscored the need for effective collision prevention measures and enhanced surveillance capabilities. Military deployments and civilian applications increasingly rely on UAVs for swift reconnaissance and various tasks. However, the surge in UAV numbers has heightened collision risks, necessitating robust collision prevention measures. In the context of the research's questions and purposes, this research aims to enhance UAV capabilities for efficient and dependable operations in dynamic environments. Specifically, it seeks to improve collision avoidance and surveillance through Charge-Coupled Device (CCD) sensors and deep learning techniques. The research leverages deep learning architectures, including You Only Look Once (YOLO), Faster Region-based Convolutional Neural Network (R-CNN), and EfficientDet, in conjunction with CCD sensors for object detection. It conducts a comprehensive comparative analysis to evaluate the performance of these architectures, with a particular focus on YOLOv5, using the UAVDT dataset. The comparative analysis reveals that YOLOv5 outperforms other architectures in terms of accuracy and speed for aerial object detection, especially when applied to the UAVDT dataset. YOLOv5 demonstrates remarkable realtime object detection capabilities, including the identification of diverse objects. While Faster R-CNN and EfficientDet models offer competitive accuracy, they require longer inference times and more training epochs to achieve comparable results. This study showcases the potential of deep learning algorithms to enhance UAV capabilities, making them more efficient and reliable in dynamic environments, thus contributing to the advancement of UAV technology. The research findings presented herein have significant implications for the ongoing development and deployment of UAVs across various sectors.
Anusha N, Md Fardeen, Chandana K, Md Abdullah, Balaram Mishra, and Vasanth K
IEEE
The paper studies the various functionality of the On-Board Computer (OBC) used in 1U CubeSat for Leo orbit. The OBC in CubeSat is responsible for controlling and managing all the subsystems of the Spacecraft. The functionality includes modes of operations of the satellite, socket programming that is used to communicate between subsystems, user datagram protocol as transportation layer protocol between different subsystems, and hardware to visualize the functional working of the satellite. The modes of operation is timer based and implemented using Java in Visual Studio (VS) code platform that acts as a real-time operating system (RTOS) of the satellite. The Nested socket programming between various subsystems was developed in Java finally the functionality of Lora was implemented through Arduino board. The packet transfer between each subsystem is analyzed using packet tracer software. The functionalities were successfully studied implemented and analyzed about the onboard computer
Naresh Alapati, N. Anusha, P Joharika, N. Jenny Jerusha, and P Tanuja
IEEE
The disruption of the brain cells that create the chemical dopamine, which enables brain cells to communicate with one another, causes Parkinson's illness. Dopamine-producing cells in the brain are what give movements control, adaptability, and fluidity. One of the most deadly and progressive nervous system illnesses that impact movement is Parkinson's. It is the second most prevalent neurological illness that impairs function, shortens lifespan, and currently lacks a treatment. Speech problems, tremor, rthymic shaking, and delayed movement impact over 90% of those who have this disease (bradykinesia). When considering the word "Parkinson's," speech traits are the major idea. Parkinson's disease can be predicted using a variety of machine learning approaches, including SVM, KNN, RF, and LR Models. The dataset is constructed using user input as well as algorithmic input. They employ the KNN Model in the current system to predict PD. The Proposed System makes use of the RF Model. Prediction is crucial, and the proposed system has a higher accuracy rate for recovering patients in the early phases. Machine learning can be used to complete this procedure.
N. Anusha
Springer Nature Singapore
Anusha NALLAPAREDDY
Politechnika Lubelska
Detection and classification of vegetation is a crucial technical task in the management of natural resources since vegetation serves as a foundation for all living things and has a significant impact on climate change such as impacting terrestrial carbon dioxide (CO2). Traditional approaches for acquiring vegetation covers such as field surveys, map interpretation, collateral and data analysis are ineffective as they are time consuming and expensive. In this paper vegetation regions are automatically detected by applying simple but effective vegetation indices Normalized Difference Vegetation Index (NDVI) and Soil Adjusted Vegetation Index (SAVI) on red(R) and near infrared (NIR) bands of Landsat-8 satellite image. Remote sensing technology makes it possible to analyze vegetation cover across wide areas in a cost-effective manner. Using remotely sensed images, the mapping of vegetation requires a number of factors, techniques, and methodologies. The rapid improvement of remote sensing technologies broadens possibilities for image sources making remotely sensed images more accessible. The dataset used in this paper is the R and NIR bands of Level-1 Tier 1 Landsat-8 optical remote sensing image acquired on 6th September 2013, is processed and made available to users on 2nd May 2017. The pre-processing involving sub-setting operation is performed using the ERDAS Imagine tool on R and NIR bands of Landsat-8 image. The NDVI and SAVI are utilized to extract vegetation features automatically by using python language. Finally by establishing a threshold, vegetation cover of the research area is detected and then classified.
N. Anusha and B. Bharathi
Elsevier BV
Anusha Nallapareddy and Bharathi Balakrishnan
Agora University of Oradea
Natural Calamities like floods cause wide-range of damage to human existence as well as substructures. For automatic extraction of flooded area in multi-temporal satellite imagery acquired by Sentinel-1 Synthetic Aperture Radar (SAR), this paper presents two neural network algorithms: Feed-Forward Neural Network, Cascade-forward back-propagation neural network. This work currently focuses on Uttar Pradesh in India, which was affected due to floods during August 2017. The two models are trained, validated and tested using MATLAB R2018b. The models are first trained using a variety of input data until the percentage of error with respect to water body detection is within an acceptable error limit. These models are then used to extract the water features effectively and to detect the flooded regions. Finally, flood area is calculated in sq. km in during flood and post-flood imagery using these algorithms. The results thus obtained are compared with that from the binary thresholding method from previous studies. The results show that the Feed- Forward Neural Network gives better accuracy than the Cascade-forward back propagation neural network. Based on the promising results, the proposed method may assist in our understanding of the role of machine learning in disaster detection.
N Anusha, M Sai Chaithanya, and Guru Jithendranath Reddy
IOP Publishing
Abstract Weather forecasting is one of the applications of science and technology, used to predict the weather condition depending on the input attributes. Most of the existing systems are implemented using statistical approaches for Support Vector Machine (SVM), which are incapable of giving the accurate prediction as they cannot capture sudden changes in weather conditions. The proposed technique uses the concept of Multi-Linear regression which can produce better results than existing methods.
N. Anusha and B. Bharathi
IEEE
Satellite imagery based change detection plays an important role in analyzing the after effects of natural disasters, detecting the changes in city limits due to rapid urbanization, updating the map database, monitoring the factors impacting agriculture, etc., The remote sensors mounted on satellites or aircrafts absorb the light reflected by the earth’s surface. The output of these sensors will be a digital image which represents the scene being perceived. In order to extract the useful information from these images, various image processing techniques need to be employed. In this paper, a detailed outline of the steps and various techniques used for detecting the changes in multi temporal remote sensing images is discussed and a case study is done by taking multi-temporal Landsat-8 images covering Hyderabad city. Image differencing method is applied in order to find the changes in the Hyderabad city limits over 2013December and 2017 December time periods.
N. Anusha and B. Bharathi
American Scientific Publishers
Prodigious flooding in the state of Uttar Pradesh, India during the month of August 2017 was induced by heavy rainfall, causing water levels in several rivers to cross the danger mark bringing normal life to a standstill. The peculiar rainfall pattern in India makes it highly vulnerable to floods. Demand for crisis information, for instance, natural disasters like severe flood events has increased. A simple but effective method is proposed in this study to find the areas that are affected due to floods, to detect the changes and for flood mapping. These indicators were derived from the Sentinel-1A Synthetic Aperture Radar (SAR) data by taking the crisis and archive images. An open flood surface can be detected easily in SAR data as it acts as a specular reflector that scatters the energy away from the sensor, causing relatively dark pixels of low backscattered SAR data. In contrast, the surrounding non-water areas usually exhibit a higher return due to surface roughness. Red, Green, Blue (RGB) composite is made for highlighting the flooded areas and for detecting changes by combining both archive and crisis images. Finally the flood map is compared with the optical imagery on the Google earth by integrating the resultant RGB composite image on the Google earth. Identification of the flood-prone areas is crucial to action the appropriate control measures in the flood-affected regions.
B. Rajalakshmi and N. Anusha
IEEE
With the increase of Android OS mobile's usage day-to-day, mobiles are getting affected with malware applications. Many antimalware's are available in the market to detect and remove these malwares from the device. But these antimalware's fails to detect the once the malware changes its form. To overcome this, we proposed a technique using SVM (Support Vector Machine) tool, which increases the malware detection strength. Each time when a new application is installed in the mobile, the permission features and API (Application Programming Interface) calls related to the application are extracted and weights are assigned to them. The weights are assigned based on their malicious nature. If the total weight exceeds the predefined threshold then it will considered as malware and reports to the user. This method can also detect even if the malware changes its form.
G. Abilash, N. Anusha, and T. Balaji
IEEE
Present days the serious issues faced by the users in cloud storage are duplication of data and access restriction to files. To solve these issues, in this paper we proposed a method which can resolve data duplication and can also assign multiple owners for a single file. The method employs an auditor who monitors for incoming request, which has been made by the owners in the group to make changes to the original data. The owner (one who created the file) only has the authority to approve or disapprove the request made by the other owners in the group. An e-mail system is introduced to notify the other owners in the group regarding the same. The file is updated only when the owner approves the request.
N. Anusha, A. Darshan Sai, and B. Srikar
IEEE
To overcome the security issues faced with the locker system now-a-days, the locker security system is proposed using IoT (Internet of Things), face recognition and OTP (One Time Password). Each time when an individual wants to access the locker, he/she should type the PIN (Personal Identification number) which is already set by them. If the PIN entered is valid, the camera connected to the locker captures and compares the image of the user using Eigen face and PCA (Principle Component Analysis) algorithms. Then the image will be matched with the database records by using Euclidean distance and Viola-Jones algorithm. Once the image gets matched One Time Password (OTP) will be generated and sent via SMS (Short Message Service) to the registered mobile using GSM (Global System for Mobile Communication) service and e-mail notification also will be sent. If the features are not matched then the log details related to the transaction will be sent to the registered number and to e-mail account. By using which the owner can easily complaint to the police officials about intruder by sending the log and location details via SMS.
M.Siva Sindhuri and N. Anusha
IEEE
In the process of analyzing the document images, the text separation from the background plays a vital role. If the quality of document image is good, then text can be separated very easily by applying simple techniques of thresholding. Whereas, in noisy images it requires very accurate analysis. Various thresholding techniques that are available are efficient and produce better results, but they are very slow in that they require threshold calculation for each and every pixel. In this paper, we overcome this problem by proposing an Otsu's method in which it follows a recursive method for segmenting an image. The experimental results show that the proposed Otsu's approach is able to extract the text very effectively from its background even if the input is a noisy image.