Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Spectral Tail Auxiliary Learning for AI-Generated Image Detection

About

As generative image models evolve rapidly, the perceptual gap between generated and real images continues to narrow, making AI-generated image detection increasingly challenging. Many existing methods exploit frequency-domain cues for detection, typically described as frequency-domain artifacts or high-frequency discrepancies. However, the specific and recurring spectral regularities remain insufficiently understood and characterized. In this paper, we systematically analyze the one-dimensional radial log-power spectra of real and generated images. We find that generated images do not necessarily exhibit higher or lower energy across the entire spectrum or high-band range. Instead, their spectra deviate from the power-law decay and show an anomalous uplift in the ultra-high-frequency tail. We term this phenomenon spectral tail uplift. We further attribute this phenomenon to nonlinear harmonic accumulation in trained generative models, suggesting that it can serve as a structural cue across generative architectures. Based on this observation, we propose Spectral Tail Auxiliary Learning (STAL), a frequency-domain auxiliary supervision framework for generalizable AI-generated image detection. STAL transfers spectral-tail cues from a tail-aware frequency teacher to a spatial detector during training, while all frequency-domain modules are discarded at inference time. Consequently, STAL introduces no inference overhead. Extensive experiments on 9 public datasets show that STAL achieves strong generalization and stability across generators, data distributions, and real-world scenarios.

Xingyi Li, Jiahui Zhang, Yiheng Li, Yun Cao, Wenhao Wang• 2026

Related benchmarks

TaskDatasetResultRank
AI-generated image detectionGenImage
Midjourney Detection Rate98.2
154
Synthetic Image DetectionForenSynths (test)
Mean Accuracy92.6
60
Synthetic Image DetectionDRCT-2M
Average Score99.8
57
AIGC DetectionAIGCDetectBenchmark--
50
AIGI DetectionAIGCDetect
B.Acc96
46
AIGI DetectionSynthWildx
DALLE3 Performance Score95.6
46
AIGI DetectionSynthbuster
B.Acc99.7
35
AIGI DetectionWildRF
FB Score98.4
23
AI-generated image detectionSynthbuster
DALL·E 2 Score99.6
23
AIGC DetectionSynthWildx
Balanced Accuracy95.2
23
Showing 10 of 15 rows

Other info

Follow for update