Project

Physics-Based Real-Time Deepfake Detection on FPGA

Built a deepfake detector on 25 interpretable image features (sensor noise, Laplacian, gradient, spectral, steganalysis, texture) instead of a heavy neural network, targeting low-power edge devices. Trained a stacked ensemble (Random Forest, Extra Trees,

• Python
• Random Forest
• Extra Trees
• XGBoost
• LightGBM
• Verilog
• Intel Cyclone V FPGA
• Quartus
• Icarus Verilog