English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deepfake videos are becoming increasingly realistic, showing subtle tamperingtraces on facial areasthat vary between frames. Consequently, many existingDeepfake detection methods struggle to detect unknown domain Deepfake videoswhile accurately locating the tampered region. To address thislimitation, wepropose Delocate, a novel Deepfake detection model that can both recognizeandlocalize unknown domain Deepfake videos. Ourmethod consists of two stagesnamed recoveringand localization. In the recovering stage, the modelrandomlymasks regions of interest (ROIs) and reconstructs real faces without tamperingtraces, resulting in a relatively good recovery effect for realfaces and a poorrecovery effect for fake faces. Inthe localization stage, the output of therecoveryphase and the forgery ground truth mask serve assupervision to guidethe forgery localization process. This process strategically emphasizes therecovery phase of fake faces with poor recovery, facilitating the localizationof tampered regions. Ourextensive experiments on four widely used benchmarkdatasets demonstrate that Delocate not onlyexcels in localizing tampered areasbut also enhances cross-domain detection performance.

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