nashid/code2image

Intelligent Code Analysis by Computer Vision Techniques and Application to Vulnerability Prediction

Code2Image: Intelligent Code Analysis by Computer Vision Techniques and Application to Vulnerability Prediction

Paper: https://arxiv.org/abs/2105.03131

Source Code Representation

• demo.py: Demonstration of AST-based image representation of source code with a sample C file (main.c)
• nodeencoding.py: Encodes AST tokens to RGB values
• treetraversal.py: Tree traversal over AST
• image_representation.py: Generates image representation of source code in numpy matrice format

Vulnerability Prediction

• readdraper.py: Read data (i.e. function source codes from the Draper dataset (https://osf.io/d45bw/)), generate image representation, rename and put them either in folder 0 (i.e. non-vulnerable) or in folder 1 (i.e. vulnerable) for each vulnerability type.
• copyfiles.sh: Creates multiple copies of samples for oversampling to deal with data imbalance
• FCNdense.py: Convolutional neural network (CNN) architecture
• generator.py: Generator for loading and processing images
• train.py: Script for training the model
• plotPR.py: Plots precision-recall curve and calculates several performance metrics

The Pretrained_Models folder contains a pretrained model for each vulnerability type