deep-tempest: HDMI Image Restoration via Deep Learning
The deep-tempest tool uses deep learning to restore images obtained by intercepting unintended electromagnetic emissions from HDMI cables.
Key Features and Benefits
- Significantly improves the quality of restored images compared to previous methods
- Reduces character recognition error rate from 90% to less than 30%
- Based on a modified version of gr-tempest
- Applies the DRUNet deep learning architecture to enhance image clarity
- Developed in Python
The program enables improved accuracy in analyzing electromagnetic monitor leaks via HDMI connections using modern neural network technologies.













