# SMART: Semantic Malware Attention Recognition Transformer

This repository contains the implementation of the SMART model for multi-class malware classification using the Maleviz dataset.

## 🔍 Abstract

SMART (Semantic Malware Attention Recognition Transformer) leverages ResNet-101 as a feature extractor, followed by Pairwise Semantic Attention (PwA) and Relation-wise Attention (RwA) modules. The model is capable of accurately classifying malware images into multiple families with high precision and F1-score.




## Dataset

This project uses the Malevis dataset available on Kaggle:

🔗 https://www.kaggle.com/datasets/your-username/malevis

unzip malevis.zip -d ./data/malevis
