eduzhai > Applied Sciences > Engineering >

Investigating the Impact of Pre-processing and Prediction Aggregation on the DeepFake Detection Task

  • Save

... pages left unread,continue reading

Document pages: 11 pages

Abstract: Recent advances in content generation technologies (widely known asDeepFakes) along with the online proliferation of manipulated media contentrender the detection of such manipulations a task of increasing importance.Even though there are many DeepFake detection methods, only a few focus on theimpact of dataset preprocessing and the aggregation of frame-level tovideo-level prediction on model performance. In this paper, we propose apre-processing step to improve the training data quality and examine its effecton the performance of DeepFake detection. We also propose and evaluate theeffect of video-level prediction aggregation approaches. Experimental resultsshow that the proposed pre-processing approach leads to considerableimprovements in the performance of detection models, and the proposedprediction aggregation scheme further boosts the detection efficiency in caseswhere there are multiple faces in a video.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×