Deep fake technology has significantly advanced the creation of synthetic images and videos, sparking widespread concerns about its potential misuse in spreading misinformation, violating privacy, and enabling identity theft. As these manipulations be-come increasingly sophisticated, the development of reliable detection methods has become a pressing necessity. This research tackles this challenge by proposing a robust deep fake detection pipeline, leveraging a custom dataset created using Roboflow. The dataset is divided into two primary classes: real and fake, with the fake class further categorized into three subtypes based on complexity: easy fake, mid fake, and hard fake. Easy fake images involve basic manipulations that are easily identifiable by the human eye, while mid fake images combine AI-generated and human-generated elements, and hard fake images are entirely AI-generated, posing significant challenges for detection. To ensure authenticity and diversity, real images were collected from personal networks and online repositories. We trained and evaluated four YOLO-based models YOLOv8, YOLOv9, YOLOv10, and YOLOv11 for the detection task. YOLOv8 emerged as the top-performing model, achieving an accuracy of 96.2% in distinguishing between real and fake images. Finally, addressing ethical considerations and developing countermeasures to mitigate the societal impact of deep fakes should remain a priority for future research.
A. Al Noman, Abdullah Al Afiq, Md. Humayun Kabir et al.· Discover Computing· 0 citations
The development of internet technology has increased the intensity of digital activities, but it has also been followed by an increase in cybersecurity threats, one of which is phishing attacks through malicious URLs. Phishing is a fraudulent method that is carried out by manipulating users through fake websites that resemble official websites to obtain sensitive information, such as usernames, passwords, and financial data. Conventional blacklist-based detection methods are considered less effective in recognizing new phishing URLs that continue to develop dynamically. Therefore, this study aims to analyze and compare the performance of various machine learning and deep learning algorithms in accurately detecting phishing URLs. The dataset used was obtained from Kaggle with a total of 11,054 data points and 31 features that represent the characteristics of phishing and legitimate URLs. The methods used include Logistic Regression, K-Nearest Neighbor, Support Vector Machine, Naive Bayes, Decision Tree, Random Forest, Gradient Boosting, CatBoost, Extreme Gradient Boosting, and Multilayer Perceptron. The research stages include data preprocessing, exploratory data analysis, data visualization, separation of training and testing data, model training, and performance evaluation using accuracy, precision, recall, and F1-score. The results showed that the Gradient Boosting algorithm provided the best performance with an accuracy of 0.974, an F1-score of 0.977, a recall of 0.994, and a precision of 0.986. The results show that the ensemble learning method is able to detect phishing URLs effectively and can be used to improve artificial intelligence-based cybersecurity systems.
