Delocate: Detection and Localization for Deepfake Videos with Randomly-Located Tampered Traces

arXiv:2401.13516v2 Announce Type: replace-cross
Abstract: Deepfake videos are becoming increasingly realistic, showing subtle tampering traces on facial areasthat vary between frames. Consequently, many existing Deepfake detection methods struggle to detect unknown domain Deepfake videos while accurately locating the tampered region. To address thislimitation, we propose Delocate, a novel Deepfake detection model that can both recognize andlocalize unknown domain Deepfake videos. Ourmethod consists of two stages named recoveringand localization. In the recovering stage, the modelrandomly masks regions of interest (ROIs) and …

address arxiv cs.cr cs.cv deepfake deepfake detection deepfake videos detect detection domain facial localization novel tampering traces videos