Jongwook Choi, Jaejun Yoo, Yonghyun Jeong, Jongwon Choi, Taehoon Kim, Seungryul Baek, Haeun Noh
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.
We introduce a deepfake video detection approach that exploits pixel-wise temporal inconsistencies, which traditional spatial frequency-based detectors often overlook. Traditional detectors represent temporal information merely by stacking spatial frequency spectra across frames, resulting in the failure to detect temporal artifacts in the pixel plane. Our approach performs a 1D Fourier transform on the time axis for each pixel, extracting features highly sensitive to temporal inconsistencies, especially in areas prone to unnatural movements. To precisely locate regions containing the temporal artifacts, we introduce an attention proposal module trained in an end-to-end manner. Additionally, our joint transformer module effectively integrates pixel-wise temporal frequency features with spatio-temporal context features, expanding the range of detectable forgery artifacts. Our framework represents a significant advancement in deepfake video detection, providing robust performance across diverse and challenging detection scenarios. contrast, frequency-based detectors transform pixel values into frequency-aware features, making it easier to identify subtle artifacts for more effective detection. While pixel-level artifacts decrease with advanced synthesis methods, frequency-level artifacts robustly appear in generated videos due to their invisible characteristics, even in high-quality outputs. While frequency-based deepfake video detectors can reveal invisible artifacts from synthesis models, the existing frequency-based detectors often overlook temporal artifacts. Video synthesis models frequently struggle with temporal inconsistency and unnatural frames. However, temporally-stacked spatial frequency spectra in existing detectors are insufficient for detecting temporal artifacts as they only consider changes in spatial artifacts and ignore pixel-wise temporal variations. Additionally, spatial frequency spectra relying on spatial This ICCV paper is the Open Access version, provided by the Computer Vision Foundation. Except for this watermark, it is identical to the accepted version; the final published version of the proceedings is available on IEEE Xplore.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2507.02398")
get_code_for_paper("2507.02398")
have("2507.02398")
Connect an agent — have() is free.