Browse Articles
Discover research articles across all indexed journals
Explainable AI for Abnormal Human Activity Detection in Surveillance Videos: A Review
This review paper examines the integration of Explainable AI (XAI) techniques into abnormal human activity detection from surveillance videos, emphasizing their significance in enhancing transparency, accountability, and trustworthiness in AI-powered surveillance systems. Through an exploration of XAI methods such as LIME, SHAP, and attention mechanisms, we discuss how these techniques provide insights into the decision-making process of AI models, enabling stakeholders to understand and interpret model predictions. Furthermore, we highlight the advantages of using XAI for debugging errors, identifying biases, and increasing trust among human operators in various security contexts. Additionally, we discuss ongoing research directions in XAI for abnormal human activity detection, including the development of advanced explanation techniques and the incorporation of user feedback into the explanation process. By fostering collaboration between humans and AI, XAI holds the potential to enhance the effectiveness and responsible deployment of surveillance systems, contributing to safer and more secure environments for all stakeholders.
Implementation of Vehicle Detection and Tracking Model Using YOLO and DeepSORT for Controlling Traffic Rule Violation
The field of computer vision has advanced significantly in the last several years, especially in the areas of object tracking and detection. This research offers a thorough analysis of the use of a vehicle tracking and detection model that makes use of DeepSORT (Deep Simple Online and Realtime Tracking) for tracking and YOLOv8 (You Only Look Once version 8) for detection. By combining these state-of-the-art methods, it will be possible to track and identify vehicles in a variety of situations in real-time, improving surveillance and traffic management as well as autonomous vehicle navigation. This project intends to develop computer vision applications in the transportation and surveillance sectors by means of a thorough examination and experimentation.
Distributed Ledger Approach BlockChain- Based Integrated EHR: A Ethereum Smart Contract Method
Blockchain, on the other hand, is a groundbreaking technology that provides a distributed and Decentralized environment in which nodes in a list of networks can connect to each other without the need for a central authority. It has the potential to overcome the limits of Electronic Health Record (EHR) management and create a more secure, Decentralized, and safer environment for exchanging EHR data. Further, blockchain is a distributed ledger on which data can be stored and shared in a cryptographically secure, validated, and mutually agreed-upon manner across all mining nodes. The blockchain stores data with a high level of integrity and robustness, and it cannot be altered. Electronic health records (EHRs) are digitally saved health records that provide information about a person’s health. EHRs are generally shared among healthcare stakeholders, and thus are susceptible to power failures, data misuse, a lack of privacy, security, and an audit trail, among other problems. The aim of our proposed framework is firstly to implement blockchain technology for EHR and secondly to provide secure storage of electronic records by defining granular access rules for the users of the proposed framework. Moreover, this framework also discusses the scalability problem faced by the blockchain technology in general via use of off-chain storage of the records. This framework provides the EHR system with the benefits of having a scalable, secure and integral blockchain-based solution.
Managing and Evaluating Data Integrity in machine Learning-Based Cyber Intrusion Detection
This study investigated the impact of data quality on the performance of intrusion detection systems. Experiments were conducted using intrusion datasets and machine learning models. The pre-trained models were less affected by data duplications and overlaps compared to classic ML models. Removing overlaps and duplicates from training data improved the pre-trained models' performance in most cases, but had adverse effects in datasets with highly similar sequences. The study also proposed a framework for model selection and data quality assurance for building high- quality intrusion detection systems. Additionally, we focus on optimizing nine hyperparameters within a 1D-CNN model, using two well-established evolutionary computation methods—genetic algorithm(GA) and particle swarm optimization(PSO).The performances of these methods are assessed using three majordatasets—UNSW-NB15,CIC- IDS2017, and NSL-KDD. The key performance metrics considered in this study include the accuracy, loss, precision, recall, and F1-score. The results demonstrate considerable improvements in all metrics across all datasets, for both GA- and PSO-optimized models, when compared to those of the original non optimized 1D-CNN model.
Smart AI Powered Auto Bill generation using Deep Learning and IOT
[1]. Smart shopping cart: In [1] this paper the author designed system for shopping mall. The system is placed in the trolleys. It consists of RFID reader and each product has RFID tag. The billing is done in smart trolley itself. Product name and its price display on LCD screen. At the cash counter the total bill relocated to Personal Computer by wireless Radio Frequency module. The disadvantage of this scheme is after completion of shopping, a key is pressed indicating the final promoting amount of the entire item, and we can’t add or remove the products. [2]. Innovative shopping cart:In [2] this paper the author designed system for shopping mall. In this paper they developed a smart way for shopping. In every item has RFID tag in its place of barcode scanner. The smart trolley involves of RFID reader LCD screen and ZigBee transmitter. The sum total cost of all the products will be added to the final bill, which will be kept in the microcontroller memory. Drawback of the system is in ZigBee there is distance barrier is occurred. [3]. Smart trolley using Arduino: In [3], this paper the author invented such a system it consists of RFID, ARM7, Display, Power supply, Switch, IR sensor pair, Barcode reader Visual Basics. The system works as the inventor use the barcode reader as well as RFID. And our shopping is done through the serial communication we transferred the total bill to the counter for printing the hardcopy of the product bill. There is also facility for the payment is credit card or any other. The drawback of the system is their uses both the reader RFID and Barcode Reader because of that the system becomes more multifaceted. [4]. RFID based advanced shopping trolley for super market: In this paper the author found such a system it consists of GSM, RFID, Automatic Billing, OTP, ZIGBEE, and PIC. In this scheme, the item can be read by the reader and the total of the Item is showed on the LED screen they categorized this paper because they added some additional features in the existing system like in this system the product weight and the name shown on display if item weight is less than stored weight then buzzer will beep.
Weed Identification Using Deep Learning in Vegetable Farming
Weed identification in vegetable plantation is more challenging than crop weed identification due to their random plant spacing. So far, little work has been found on identifying weeds in vegetable plantation. Traditional methods of crop weed identification used to be mainly focused on identifying weed directly; however, there is a large variation in weed species. This paper proposes a new method in a contrary way, which combines deep learning and image processing technology. Firstly, a trained CenterNet model was used to detect vegetables and draw bounding boxes around them. Afterwards, the remaining green objects falling out of bounding boxes were considered as weeds. In this way, the models focus on identifying only the vegetables and thus avoid handling various weed species. Furthermore, this strategy can largely reduce the size of training image dataset as well as the complexity of weed detection, thereby enhancing the weed identification performance and accuracy. To extract weeds from the background, a colour index-based segmentation was performed utilizing image processing. The employed colour index was determined and evaluated through Genetic Algorithms (GAs) according to Bayesian classification error. During the field test, the trained CenterNet model achieved a precision of 95.6%, a recall of 95.0%, and a F1 score of 0.953, respectively. These experiment results demonstrate the feasibility of using the proposed method for the ground-based weed identification in vegetable plantation.
Predictive Analysis and Discerning of Polycystic Ovary Syndrome in Women using Machine Learning Computations
This paper focuses on the data-driven diagnosis of polycystic ovary syndrome (PCOS) in women. For this, machine learning algorithms are applied to a dataset freely available in Kaggle repository. This dataset has 43 attributes of 541 women, among which 177 are patients of PCOS disease. Firstly, univariate feature selection algorithm is applied to find the best features that can predict PCOS. The ranking of the attributes is computed and it is found that the most important attribute is the ratio of Follicle stimulating hormone (FSH) and Luteinizing hormone (LH), Next, holdout and cross validation methods are applied to the dataset to separate the training and testing data. A number of classifiers such as gradient boosting, random forest, logistic regression, and hybrid random forest and logistic regression (RFLR) are applied to the attributed are good enough to predict the PCOS disease
Revolutionizing Deepfake Detection and Realtime Video Vision with CNN-based Deep Learning Model
The rapid advancement of deep learning models that can generate and synthesis hyper-realistic videos known as Deepfakes and their ease of access have raised concern on possible malicious intent use. Deep learning techniques can now generate faces, swap faces between two subjects in a video, alter facial expressions, change gender, and alter facial features, to list a few. These powerful video manipulation methods have potential use in many fields. However, they also pose a looming threat to everyone if used for harmful purposes such as identity theft, phishing, and scam. In this work, we propose a Convolutional Vision Transformer for the detection of Deepfakes. The Convolutional Vision Transformer: Convolutional Neural Network (CNN) and Vision Transformer (ViT). The CNN extracts learnable features while the ViT takes in the learned features as input and categorizes them using an attention mechanism. We trained our model on the DeepFake Detection Challenge Dataset (DFDC) and have achieved 91.5 percent accuracy, an AUC value of 0.91, and a loss value of 0.32. Our contribution is that we have added a CNN module to the ViT architecture and have achieved a competitive result on the DFDC dataset.
Transforming the University HR on Boarding Process with RPA
The HR Onboarding Assistant project employs Robotic Process Automation (RPA) to enhance the efficiency of on boarding in education institutions. This initiative utilizes a well-structured RPA system with bots, data integration, security measures, and user-friendly interfaces to automate tasks, reduce errors, and improve the onboarding experience. It also supports ongoing optimization and data-driven decision-making for colleges and universities. This initiative addresses these challenges by introducing a meticulously designed RPA architecture. The architecture encompasses a suite of RPA bots, an orchestration layer, and compliance measures. Each component collaborates to automate specific tasks, thereby reducing manual workload, eliminating human errors, and accelerating the onboarding timeline. The project offers a comprehensive solution that includes a user-friendly interface for HR professionals, real-time notifications. Furthermore, the architecture accommodates ongoing optimization, and continuous monitoring. It is designed to foster a culture of data-driven decision - making, ensuring that the onboarding process evolves and adapts to the institution's needs.
Load Balancing using Genetic Algorithm with Cloud SIM Simulator in Open Source
Clouds are highly configurable infrastructure that offer software and platforms as a service, enabling users to subscribe for the services they need using a pay-as-you-go model. The popularity of cloud computing is growing due to its straightforward, service-oriented architecture. Every day, more and more people are using the cloud. Cloud computing is typically built on data centres with the capacity to support huge numbers of users. The ability of clouds to manage loads determines how reliable they are in order to address this issue; clouds need to include a load balancing system. In cloud computing, load balancing will help clouds become more capable and reliable by boosting their capacity.
Computer Vision-Based Mobile Application for Visually Impaired
In our society, there's a growing need for awareness and support, especially for individuals facing visual impairments. This endeavor revolves around creating a specialized mobile application precisely catering to their unique needs. Harnessing the capabilities of computer vision and deep learning, the app functions as a versatile aid, offering real-time support and enhancing accessibility for its users. Through the Smartphone’s camera, the application can identify and describe various objects, allowing users to gain a better understanding of their immediate environment. Furthermore, the integration of optical character recognition (OCR) technology enables the app to convert printed text into audio, providing users with access to a wide range of printed materials. The navigation assistance feature utilizes advanced algorithms to detect obstacles and provide users with alerts, ensuring safe and efficient navigation.
Predicting Indian GDP with Machine Learning: A Comparison of Regression Models
In recent years, the application of machine learning algorithms in economic forecasting has gained significant traction. This study focuses on predicting the Gross Domestic Product (GDP) of India by harnessing advanced machine learning techniques. We assembled a comprehensive dataset incorporating time series analysis and inflation rates from diverse sources. Our investigation involved a comparative analysis employing both linear and polynomial regression methods to discern the most accurate predictive model. Our findings underscore the superiority of the polynomial regression model, which excelled in capturing non-linear relationships between independent variables and GDP. Specifically, the polynomial regression model achieved an impressive prediction accuracy rate of surpassing the accuracy achieved by the linear regression model. This study underscores the critical role of advanced machine learning algorithms in economic forecasting. By emphasizing the significance of high-quality datasets and the application of techniques like polynomial regression, we illustrate how these factors can substantially enhance the precision of economic forecasts. These insights hold substantial implications for policy makers and businesses alike, aiding informed decision-making and effective economic policy development. This study serves as a valuable reference for future research, highlighting the importance of advanced machine learning methods and robust data sources in economic forecasting.
Boosting Network Security: A comparative Analysis of Deep Learning Techniques for Intrusion Detection
This paper addresses demanding situations in Intrusion Detection Systems (IDS) by way of combining the Adaptive Synthetic Sampling (ADASYN) method with a break up-primarily based Resnet framework. ADASYN balances sample distribution, overcoming biases in the direction of large samples. Our method extracts multiscale functions, reduces interchannel redundancy, and enhances the version's capability using a smooth hobby operation. We recommend a Residual Neural Network (ResNN) version for intrusion detection, displaying massive enhancements in recognition accuracy and execution performance. Ongoing work targets to optimize performance further, highlighting the capacity for extra strong IDS solutions.
Gupta-Yantra : Spying Robot for Defence
Recently, with rising public security awareness, the demand for an efficient surveillance system has become highly significant. But, several monitoring systems developed have certain drawbacks limiting their applications. Therefore, a refined surveillance system has been proposed which could be utilized in many fields, especially in border areas for intrusion detection, to reduce the crime rates and to serve tremendous purposes. The project focuses on building a spying robot for maneuvering and monitoring in perilous environments. The module includes Raspberry Pi interfaced with DC motors, GPS, Pi camera, Temperature sensor, Fire sensor and ultrasonic. Meanwhile, the motion of the robot could be controlled by a web application through Wi-Fi connectivity. This robot is programmed to stream the live data over a virtual network and update the captured images along with the time stamp and location at certain instances to the cloud. Hence, it can transmit real-time videos and pictures with temperature measured.
A Deep Learning Based Soil Type Classification and Recommendation of Crops to the Soil Type
Accurate classification of soil surface texture plays a pivotal role in agriculture, environmental management, and land-use planning. However, the classification of soil textures from RGB images obtained under uncontrolled field conditions remains a challenging task due to the inherent complexities of natural environments. In this study, we propose a novel approach for Soil Surface Texture Classification using Convolutional Neural Networks to address these challenges.
Securing IoT Healthcare Data Using Machine Learning Techniques
The internet of Things has become a crucial technology, enabling the interconnection of everyday devices like home appliances, medical equipment and vehicles. However, the sensitive data handled by IoT systems, particularly in the healthcare sector, requires robust security measures to prevent tampering or theft. Detecting security threats in IoT environments requires sophisticated techniques. Different machine learning algorithms are used to identify the attacks in IoT network traffic and predict snooping behavior based on unidentified patterns. This paper proposed a method to identify network traffic attacks by employing different strategies utilizing the NT-ToN-IoT dataset. The classifier utilized Naïve Bayes(NB), Random Forest(RF), Support Vector Machine, Artificial Neural Network. These algorithms favoured over centralized methods to construct streamlined security system for IoT devices. Prior to analysis the dataset underwent preprocessing to eliminate irrelevant or missing data. Following this, a feature engineering approach is applied to extract crucial features. The results from applying each classifier demonstrated a significant maximum classification accuracy of 98% achieved by the RF model.
Enhancing Signature Forgery Detection System Using CNN-SVM
Handwritten signatures are widely used as a means of personal identification and authentication. Many documents like bank cheques and legal transactions require signature verification. But considering a large number of documents, it is a very difficult and time-consuming task. Therefore, ensuring the necessity for a robust automatic signature verification tool that aims to reduce fraud in all related financial transaction sectors. The current visual verification depends mainly on the experience, mood, and working environment of the verifier which ultimately wastes both time and money. Moreover, it is difficult for the eyes of any experts to precisely verify the ratios between lines and angles of a genuine signature to a fraud signature. Therefore, we propose an automatic signature verification technique using the recent advances in image processing and machine learning.
Material Property Prediction Using Machine Learning
The ability to accurately predict the properties of materials is crucial for numerous applications across various industries, including materials science, engineering, and manufacturing. With the advent of machine learning (ML) techniques, researchers have gained powerful tools to model and predict material properties based on their composition, structure, and processing conditions. This review paper provides a comprehensive overview of material property prediction using machine learning. It covers the historical development, available ML models, recent trends, and prospects in this rapidly evolving domain.
Generative Adversarial Networks for Malware Detection in Cloud Computing Environments
In recent years, the proliferation of sophisticated malware threats has necessitated the development of advanced detection techniques to safeguard computer systems and networks. Generative Adversarial Networks (GANs), a class of deep learning models comprising a generator and a discriminator trained in an adversarial manner, have emerged as a promising tool for improving malware detection capabilities. Traditional signature-based malware detection struggles to keep pace with the ever-evolving threat landscape in cloud computing environments. This paper proposes a novel approach utilizing Generative Adversarial Networks (GANs) for enhanced malware detection. The GAN architecture generates realistic malware samples, refining the discriminator's ability to differentiate between legitimate and malicious software. This fosters a model adept at identifying zero-day attacks and unseen malware variants. Evaluation in a cloud environment assesses detection accuracy, efficiency, and generalizability. Findings underscore GANs' potential to enhance cloud-based malware detection, securing the future of cloud computing.
Prediction of Alzheimer’s Disease with Retinal Images Using Deep Learning
Alzheimer's disease (AD) is a leading cause of mortality worldwide. Early detection and accurate diagnosis of AD are crucial for effective intervention and improved patient outcomes. Retinal imaging has emerged as a non-invasive and cost-effective technique for AD prediction. This study aims to develop a deep learning model using convolutional neural networks (CNNs) architecture to predict AD from retinal images. The proposed model leverages the capabilities of CNNs to automatically learn relevant features from retinal images.