Exposing DeepFakes via Hyperspectral Domain Mapping

TL;DR

This paper introduces HSI-Detect, a novel hyperspectral domain mapping approach that reconstructs 31-channel images from RGB inputs to improve DeepFake detection by revealing subtle spectral artifacts.

Contribution

The paper presents a new two-stage pipeline that reconstructs hyperspectral images from RGB inputs, enhancing DeepFake detection beyond traditional RGB-based methods.

Findings

HSI-Detect outperforms RGB-only baselines on FaceForensics++ dataset.

Spectral domain mapping amplifies manipulation artifacts in specific frequency bands.

Reconstructing hyperspectral images improves detection accuracy of DeepFakes.

Abstract

Modern generative and diffusion models produce highly realistic images that can mislead human perception and even sophisticated automated detection systems. Most detection methods operate in RGB space and thus analyze only three spectral channels. We propose HSI-Detect, a two-stage pipeline that reconstructs a 31-channel hyperspectral image from a standard RGB input and performs detection in the hyperspectral domain. Expanding the input representation into denser spectral bands amplifies manipulation artifacts that are often weak or invisible in the RGB domain, particularly in specific frequency bands. We evaluate HSI-Detect across FaceForensics++ dataset and show the consistent improvements over RGB-only baselines, illustrating the promise of spectral-domain mapping for Deepfake detection.