VisionMalNet: A CNN-Based Image Transformation Approach for Android Malware Detection

Mobile devices running the Android operating system have become essential in contemporary life, resulting in a substantial increase in malicious software designed to target the platform. This paper introduces a new method that combines computer vision methods with conventional cybersecurity techniques to improve the identification and prevention of Android malware. We propose a novel algorithm called VisionMalNet, which utilizes image-based features extracted from application behaviors, permissions, and code structures. Unlike signature-based or heuristic approaches, VisionMalNet translates malware attributes into visual formats, thereby facilitating sophisticated pattern detection via convolutional neural networks (CNNs). Experimental findings indicate a substantial increase in detection precision, resulting in a success rate of 97.6%, surpassing the performance of current pioneering methods. This study offers a thorough and practical approach to counteracting increasingly sophisticated malware threats, closing the divide between cybersecurity and computer vision.