Eyeballer uses a convolutional neural network to analyze pentest screenshots and identify potentially vulnerable or interesting web targets.
Convolutional neural network for analyzing pentest screenshots
Eyeballer is designed for penetration testers and red teamers conducting large-scale network assessments who need to quickly sift through thousands of web screenshots to find valuable targets. By classifying screenshots into categories like login pages, old-looking sites, and parked domains, it helps prioritize attack surfaces efficiently.
Setting up GPU support requires additional hardware compatibility and driver installation beyond the scope of this README. For best results, screenshots should be taken with tools like EyeWitness or GoWitness and resized to 224x224 pixels. Pretrained weights are available to use immediately without retraining.
Install required Python packages with `sudo pip3 install -r requirements.txt`
For GPU support, install packages with `sudo pip3 install -r requirements-gpu.txt`
Download pretrained weights from the GitHub releases section
Download training data from https://www.kaggle.com/altf42600/pentest-screensots
Place the `images/` folder and `labels.csv` file into the root Eyeballer directorysudo pip3 install -r requirements.txt
Installs the necessary Python dependencies for CPU-only usage
sudo pip3 install -r requirements-gpu.txt
Installs the necessary Python dependencies for GPU support