This paper presents the development and evaluation of the Real-Time Traffic Sign Detection Model (RTTSDM), lever-aging the YOLOv4 framework with a CSPDarknet53 backbone, trained on the German Traffic Sign Recognition Benchmark (GTSRB) dataset. The RTTSDM exhibits significant advancements in traffic sign detection, showing substantial improvements across key performance metrics including average Intersection over Union (IoU), precision, recall, F1 Score, and mean Average Precision (mAP) throughout training. Notably, the model achieves a peak mAP of 0.9685 at iteration 4000, reflecting its superior ability to accurately identify and classify traffic signs. The model demonstrates robust performance under various lighting conditions, effectively detecting prohibitory, danger, mandatory, and priority signs with high accuracy. The RTTSDM shows strong potential to improve road safety and traffic management through its precise and reliable traffic sign detection capabilities.

The challenge of evaluating deep learning-based object detection models in complex traffic scenarios, characterized by changing weather and lighting conditions, is addressed in this study. Real-world testing proves time and cost-intensive, leading to the proposal of a Video Frame Feeding (VFF) approach as a solution. The proposed Video Frame Feeding approach acts as a bridge between object detection models and simulated environments, enabling the generation of realistic scenarios. Leveraging the CarMaker (CM) tool to simulate realistic scenarios, the framework utilizes a virtual camera to capture the simulated environment and feed video frames to an object identification model. The VFF algorithm, with automated validation using simulated ground truth data, enhances detection accuracy to over 95% at 30 frames per second within 130 meters. Employing the You Only Look Once (YOLO) version 4 and the German Traffic Sign Recognition Benchmark dataset, the study assesses a traffic signboard identification model across various climatic conditions. Notably, the VFF algorithm improves accuracy by 2% to 5% in challenging scenarios like foggy days and nights. This innovative approach not only identifies object detection issues efficiently but also offers a versatile solution applicable to any object detection model, promising improved dataset quality and robustness for enhanced model performance.

Residential burglary is a severe crime that affects millions of residents each year. It is critical to analyze patterns of human behavior in surveillance video data and discover suspicious actions to avoid and deter this crime. However, the complexity and diversity of human behavior make this undertaking challenging. In this research, we provide a unique system that uses advanced deep learning algorithms to detect abnormal actions that signal burglar attempts at time intervals. We use the state-of-the-art object identification method YOLOv7 to detect an object and its trajectories in video data. We additionally extract the time intervals between the object's movements and utilize them as a categorization feature. Based on the COCO dataset and three videos of distinct scenarios: abnormal activity, burglary attempt, and false alarm, we evaluate our system with earlier versions of YOLO and additional detection methods. On the COCO dataset, our method outperforms all other models with a precision of 93.9

Autonomous driving faces a significant challenge in recognizing traffic signs in adverse weather conditions. This research aims to address this challenge while also reducing the cost of on-road testing and validation. Current solutions for detecting traffic signs under various driving and weather conditions are deemed ineffective and costly. To overcome these limitations, the proposed method leverages six different weather scenarios to utilize real-time sensory inputs. By integrating YoLov5 vision with a virtual simulation setup, the approach can analyze these inputs and identify traffic signs in a simulated environment using real-captured videos. The model undergoes rigorous testing and validation in a controlled environment, employing a road model and vehicle model to simulate the six different weather conditions. The results demonstrate the success of the approach across all testing conditions, with YoLov5 vision exhibiting a significantly improved level of accuracy in detecting traffic signs, showing an average improvement of 5.56% for day, 7.32% for foggy day, 5.88% for cloudy, 4.55% for dusk, 6.25% for foggy night, and 2.17% for night conditions. Additionally, the research findings highlight the cost-effectiveness of the proposed method compared to existing solutions, emphasizing its potential for enhancing the safety and efficiency of autonomous driving systems.

In image analysis and processing, image segmentation is one of the most important functions. The outcomes of segmentation have such significance on all subsequent image analysis operations, covering object tracking and description, feature measurement, and even higher-level tasks like object recognition. Malicious code detection is becoming increasingly significant, and current models must be improved. Hence forth the Image segmentation in the field of Malware image classification is a significant task. The sectional structure or region of interest must be identified and extracted during the segmentation process so that it can be evaluated independently. There are various reviews stating the traditional approach of image segmentation in various fields. The necessity of image segmentation in malicious image is extracting data for classification using CNN is discussed. In this work we use malimg_paper_dataset_imgs with 9,339 malware images. Various segmentation techniques were used to enhance the malware images. Those enhanced malicious image were applied in CNN architecture and Mal_CNN for classification and a comparative result is been discussed. The malicious images in dataset after incorporating segmentation have achieved 95% of accuracy in CNN architecture and 97% with Mal_CNN.

Artificial Intelligence (AI) is a technique that incorporates human intelligence into mundane activities. And there is no question that AI is significantly affecting security and surveillance. Although relying on numerous resources, finding answers, and implementing technology for decades, our security and surveillance systems still have flaws. In every country around the globe, the use of video security and surveillance is becoming more widespread. Nonetheless, a wide range of businesses has made use of it, including hospitals, universities, and warehouses. Yet, as people are limited in their ability to vigilantly monitor live video streams, deep learning was developed to better fill the position. Unfortunately, there are other problems with images in the real world, including jitter or blurring caused by rotating objects, noise, and sharpness concerns. As a result, deep learning technology for surveillance has considerably improved in recent years. The main objective of this study is to detect burglars using deep learning technology. This system aims to take video surveillance of the residential environment as input and pass it into the Yolo model to increase the speed and accuracy of the system to detect burglars in the residential. This system mainly concentrates on object detection.

We present an extensive driver assistance system capable of executing two essential tasks. The first module is used in assisting the driver with road safety alerts (RSA); it scans the road environment and detects any significant entities including but not limited to vehicles, pedestrians, and traffic lights. Further, alerts are issued when the estimated physical distance from detected entities is less than a set threshold. For this module, we also propose the usage of a compute-accelerated Swin Transformer model and evaluate its efficacy against other state-of-the-art models by considering relevant metrics like inference time and mAP. The second module pertains to driver alertness detection (DAD) for identifying signs of fatigue. It scans the driver’s face and monitors a live video feed to ensure that the driver shows no signs of micro-sleep. When either module detects a behavioural anomaly, it will alert the driver with text-based messages and non-disruptive audio messages. We propose such a state-of-the-art safety system being integrated into the advanced driver assistance systems (ADAS’s) seen in modern vehicles.

Raw data gets transformed into features representing the problem in a more improved manner. The problem gets represented in predictive models and in turn the accuracy is at a higher rate on the data. With feature engineering, the input data is turned into the machine learning algorithms, using the domain knowledge. Various feature extraction techniques like color, texture, shape, position and edge are worked upon a gray scale image to analyze the malware and its features. The malware image being a gray scale image, the featuring methods are performed to work on the malware. An irrelevant or partially relevant model feature may depreciate the model performance. When a model is first designed, feature selection and data cleaning are the foremost steps to be attended to. The feature selection models that are in practice are Univariate Selection, Feature Importance, Heatmap. These models can be utilized to improve the result on a model when applied.

Federation Payment Tree, a new Off-chain with zero-knowledge hash time lock commitment setup is proposed in this paper. The security of blockchain is based on consensus protocols that delay when number of concurrent transactions processed in given throughput framework. The scalability of blockchain is the ability to perform support increasing workload transaction. The FP-Tree provides zero knowledge hash lock commitment connect with off-chain protocols by using the payment channel, which enables execution of off-chain protocol that allows interaction between the parties without involving the consensus protocol. It allows to make payment across an authorization path of payment channel. Such a payment tree requires two commitment scheme is Timelock and Fundlock, each party lock fund for a time period. The main challenges we faced in this paper is that the computational power, storage and cryptography. Furthermore, we discussed many attacks on off-chain payment channel that allows a malicious adversary to make fund lose. The FP-Tree supports multi-parti computation (MPC) merging transactions into single hash value in payment tree. We enable the parties to generate single hash value by consumes both less than 0 (log2 N ) and space less than 0 (log2 N ) time combine element over length of single hash. The results were discussed in this paper and efficiency of FP-Tree is well suited for the blockchain technology. We achieved the accuracy of 60.2% in federated payment tree when compared with the proof of work and proof of authority.

Blockchain is the upcoming technology in creating the Dapp for different types of use cases. The transactions in the blockchain technology are transparent to all the nodes in the decentralized network. Most of the applications were built and trying to implement it with the Ethereum platform in blockchain for the security purpose. The nodes in the distributed network shares all the data to the other nodes without any modification of their data in the blockchain technology. The transactions done by the nodes are trusted through the digital signatures. In this paper, we built a decentralized web application for e-commerce by using the blockchain Ethereum platform. Our application is deployed by using the smart contracts, which are written in the solidity language and the front-end and back-end process are explained. All these data about the e-commerce details are stored into the blocks in blockchain. The Web application for online shopping will be secured and trust worthy for the nodes in the decentralized network. The other online shopping platforms like amazon are trying to deploy the blockchain for security of data and user’s convenient. The issues like storage, scalability, and others can be considered for the future work.

The upcoming technology in creating the distributed applications for different use cases in real world applications are Blockchain technologies. The application may be private or public depending upon the use cases. Most of the applications are built by using the Ethereum platform in blockchain for use of security purpose. The nodes in the distributed network shares all the data to the other nodes without any modification of the data in the blockchain technology. The transactions done by the nodes were trusted through the digital signatures. Here, no central control or the central system will be there to keep track of other nodes data. In this paper, we built the POA network by using the Ethereum blockchain platform. The network is called the Proof of Authority and its main purpose is to deploy the contracts especially with authority in the Ethereum blockchain platform. We also explained about the ERC20 tokens, the working process of using the POA network and the codes for connecting with the web page.

Surveillance is an observation of a place, large areas, behavior, or a variety of activities to acquire information, influence, manage, or guide it. When people talk about surveillance solutions, the growing demand for large area monitoring becomes one of the key trends in the security industry. Surveillance video is used in real-time to watch known threats. Suspicious activities through surveillance video are a major topic in image processing and deep learning research. Residential area security is very much important to people nowadays. The proposed system is concerned with the development of a surveillance video framework in the residential area to detect any type of suspicious robbery activity. This system makes effective use of deep learning techniques of yolo, this includes techniques like object detection and eventually identifying the actions required to prevent robberies. Surveillance cameras are used here to remotely monitor a residential area or building by transmitting recorded images or videos to a control station to thwart suspicious activities. As a result, deep learning techniques are employed to achieve outstanding detection of suspicious actions that yielded positive results.

In this paper, the distributed decentralized network with the nodes at any place can access the data behaviour intention performance is analysed and explained in detail. The bitcoin blockchain is elementary of several crypto-currencies. The blockchain technology holds principal assurance for a difference of enterprises, including financial services, cryptocurrency exchange, laundering of anti-money tracking system, unique identity security, healthcare, real estate, supply chain, and logistics monitoring. Results showed that blockchain is considerably new technology and the nodes are sharing their data in the network without any trust in other nodes. Furthermore, key aspects of using this technology, the users in the crypto world buy, sell, or exchange the crypto-coins in the blockchain network. Many of the protocols are there to exchange cryptocurrencies. The main contributions of this work liesbased on the protocol named uniswap protocol, which helps the users to exchange the crypto-coins. Uniswap protocol is defined to be the automated liquidity protocol provision on the blockchain Ethereum platform. With the uniswap, a user can buy or sell the ERC20 tokens in the decentralized distributed network by using an Ethereum smart contract. This paper will help the users in the blockchain network to useuniswap protocol as a trustworthy and efficient way of exchanging the crypto-coins or the tokens in the blockchain network.

Wind is a clean source of energy which is naturally available and has been replacing the harmful non-perennial conventional power plants such as Thermal, Hydro, Nuclear etc., The innovation in design has gradually reduced the cost to set up a wind power plant. This has increased the penetration of wind energy in the Distribution Network (DN). Being a seasonal power generation, wind energy integration in the distribution network has created many challenging issues like voltage stability, system reliability and power system network stability. Hence a thorough analysis of wind energy integration is needed to make the power system network more stable. This paper describes the dynamic behavior of the distribution system while integrating wind energy through time series analysis. The load flow is performed using Newton-Raphson method, with a variable load profile. Time series analysis is done which gives the performance characteristics of the system which is validated in a practical IEEE 123 test feeder.

Malware is computer infections which are malicious programs readily created by hackers or programmers to create extensive damage. Even though various antivirus programs are created to prevent the file from various usage, new malware versions and other computer infection programs are developed daily. Security researchers are working tirelessly to achieve Zero-day malware. Malware detections are done using antivirus software, automated malware systems and an online malware analysis system. Even though open-source analysis systems like cuckoo sandbox are available, security researchers face a tough time detecting and analyzing new malware variants. Deep Learning is an emerging subset of Artificial Intelligence, which deals with many complex problems and states the most accurate expected output. In this chapter, major classifications of malware families and Deep Learning algorithms will be discussed. In recent security trends, there have come a lot more differences between existing malware and new malware. This malware tends to cause a lot of damage to the system, resulting in a data leak. Though many studies detect and analyze malware, the current malware is getting complicated day by day. The Deep Learning methods analyze and rectify the malware and the affected data to overcome these hazards.

Choosing the best marketing strategies and understanding customer behaviour is what drives many companies to resort to analysing these behaviours. There are several approaches and perspectives to reach a deep understanding. In this work customers were categorised based on their buying pattern taking into account the loyalty score which was calculated using the RFM model. Mainly it is represented by three factors: recency, frequency and monetary. After performing the categorisation process, several models are trained and used to predict the cluster of the customers. In this paper, the proposed models and their results will be discussed in order to reach the best model. We also show that after discussing the models, it was found that the best model was the gradient boosted decision tree that achieved a prediction accuracy of 97% in the testing phase. By relying on this work, it was found that it is possible to classify and predict the customers' class according to their buying pattern and the level of their loyalty to the store.

Any malicious code or program that affects the system or the software program is, in general, described as malware. This malware is usually harmful if not found on the right note. With evolving technologies, this malware tends to grow in different forms as they are created to steal information or make any potential hazard to the user. Machine learning comes on the line when detecting malware with various machine algorithms like random forest, K-nearest neighbor, decision tree and SVM. The malware dataset is classified as either malicious or benign. This paper unfolds the algorithms’ efficiency by analyzing the hash codes and finds the most efficient algorithm for this particular dataset. Efficiency includes performance metrics. Thus, when the hash codes are fed into the mechanism, it delivers the performance metrics. After finding the perfect algorithm for the dataset, it is used to train the dataset to get the best results.

The departments of Agriculture are in need of tracking and updating all the details of food and the agricultural industry. This data analysis is mandatory to have the latest updated information for the authorities who decide on the market value and the prediction of various products. Having complex, high dimensional information and therefore a progressively perplexing preparing chain as a link of various calculations, up to now it was about difficult to discover what occurred in the classification procedure and which parts of the first information were utilized. The proposed strategy utilizes the Artificial Intelligence calculation using Support Vector Machine for the grouping and development of information bazaar. It empowers a representation of the total handling chain in the information space and the preprocessing to convert into semi structured data. The classifier used defined the data with distinct facts and dimensions to enhance the prediction of data analysis in the food and agricultural sector.

Lately, the usage of Information system and strategic tools in the domain of medical science is constantly growing. The liver is an Exocrine Gland that helps in the digestion process, especially for fats along with altering the pH of food due to its alkaline nature. One of the most common signs for the majority of liver disease is hyperbilirubinemia which is very difficult to identify at an early stage. While certain diseases such as obstructive Jaundice and acute viral hepatitis present themselves with an early rise of bilirubin along with yellowish discoloration of the skin, many other diseases don’t usually present themselves with an early rise of serum bilirubin or skin discoloration because of which sometimes liver disease are overlooked or misdiagnosed in primary level. However, serum bilirubin is not the only way to diagnose liver disease because it is not specific. The most specific way to diagnose liver disease is by liver function test. With the help of the detection of the enzyme level, we can identify and confirm the presence of liver disease and intensity of liver damage when coupled with suitable imaging modalities like Ultrasound, CT scan, or MRI scan. The dataset contains patient parameters such as Age, Sex, Total Proteins, Alkaline Phosphatase, Alanine Phosphatase, Total Proteins, Total Bilirubin, Albumin, Albumin and Globulin Ratio, Direct Bilirubin, and the Result. We are using Binary Classification which is basically classifying the element of given set into two given sets i.e., Patient suffering from Liver disease or not. We will try to use an Ensemble Based Approach to find the best prediction accuracy.

Massive growth in the big data makes difficult to analyse and retrieve the useful information from the set of available data’s. Statistical analysis: Existing approaches cannot guarantee an efficient retrieval of data from the database. In the existing work stratified sampling is used to partition the tables in terms of static variables. However k means clustering algorithm cannot guarantees an efficient retrieval where the choosing centroid in the large volume of data would be difficult. And less knowledge about the static variable might leads to the less efficient partitioning of tables. Findings: This problem is overcome in the proposed methodology by introducing the FCM clustering instead of k means clustering which can cluster the large volume of data which are similar in nature. Stratification problem is overcome by introducing the post stratification approach which will leads to efficient selection of static variable. Improvements: This methodology leads to an efficient retrieval process in terms of user query within less time and more accuracy.