Exposing DeepFakes via Hyperspectral Domain Mapping

View on arXiv ← Back to list

Authors: Aditya Mehta, Swarnim Chaudhary, Pratik Narang, Jagat Sesh Challa

Published: 2025-11-13 06:25:44+00:00

Comment: Accepted at AAAI 2026 Student Abstract

AI Summary

This paper introduces HSI-Detect, a two-stage pipeline for deepfake detection that reconstructs a 31-channel hyperspectral image from a standard RGB input. By expanding the input representation into denser spectral bands, HSI-Detect amplifies manipulation artifacts that are often weak or invisible in the RGB domain. The approach demonstrates consistent improvements over RGB-only baselines on the FaceForensics++ dataset, highlighting the promise of spectral-domain mapping for deepfake detection.

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.

Key findings
HSI-Detect consistently outperforms prior RGB-only baseline methods (ViT, RECCE, MoE-FFD) across multiple unseen manipulation types, achieving the highest average AUC. The gains are particularly significant on DeepFakes and FaceSwap manipulations, demonstrating that hyperspectral cues provide strong discriminative power for more robust and generalizable deepfake detection.
Approach
The HSI-Detect pipeline consists of two stages. First, it uses the MST++ model to reconstruct a 31-channel hyperspectral image from an RGB input. Second, an enhanced version of the UCF disentanglement framework, incorporating an encoder, decoder, and two classification heads, analyzes this hyperspectral representation to detect whether the input is real or fake.
Datasets
FaceForensics++
Model(s)
UNKNOWN
Author countries
India
← Previous
Next →