Mal Class A Deep Learning Approach for Automatic Classification of Malware Images

This project presents a deep learning-based system for the automated binary classification of malware as either benign or malicious through image-based analysis. The methodology involves converting malware binaries into grayscale images, which are then processed by a suite of advanced convolutional and transformer-based neural network architectures. The system comparatively evaluates the performance of multiple models, including MobileNet, DenseNet, EfficientNet+Swin Transformer, and ConvNeXt+Tiny DeiT, to identify the most effective architecture for this task. A user-friendly web application is developed using HTML, CSS, and the Flask framework to operationalize the selected model. This interface allows users to register, log in securely, and upload suspected malware images for instantaneous classification. The deployed model analyzes the uploaded image and returns a clear detection result, aiding in rapid and accessible malware screening. The project thus integrates state-of-the-art deep learning techniques with a practical application interface to create an effective tool for automated malware visual analysis.

Keywords:
Malware Classification, Deep Learning, Image-based Analysis, Binary Classification, Convolutional Neural Networks (CNN), Transformers, Flask Web Application, Benign vs. Malicious Detection, Model Comparison, Security.