AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

OpenForensics: Large-Scale Challenging Dataset for Multi-Face Forgery Detection and Segmentation In-the-Wild

Trung-Nghia Le, Huy H. Nguyen, Junichi Yamagishi, Isao Echizen

ICCV 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Industrial anomaly detection · 110 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use OpenForensics instead of Celeb-DF as the source of forged faceshuman viewers · image realism user studymean opinion score: 4.0 vs 3.2Improved
use OpenForensics instead of Celeb-DF as the source of forged faceshuman viewers · face forgery classificationhuman accuracy: 59.7% vs 68.7%Hurt
use OpenForensics instead of FaceForensics++human viewers · face forgery classificationhuman accuracy: 59.7% vs 82.0%Hurt
increase the number of forged faces in an imagehuman participants · multi-face forgery detectionhuman detection accuracy: 86% -> 0%Hurt
use BlendMask instead of MaskRCNNdetector with FPN-ResNet50 · OpenForensics detection (test-dev)AP: 79.2 -> 87.0Improved
use BlendMask instead of YOLACTdetector with FPN-ResNet50 · OpenForensics detection (test-dev)AP: 68.1 -> 87.0Improved
evaluate on unseen test-challenge images instead of test-devBlendMask · OpenForensics detectionAP: 87.0 -> 53.9Hurt
evaluate on unseen test-challenge images instead of test-devMaskRCNN · OpenForensics detectionAP: 79.2 -> 42.1Hurt
use YOLACT++ instead of MaskRCNN on unseen imagesdetector with FPN-ResNet50 · OpenForensics detection (test-challenge)AP: 42.1 -> 53.7Improved
use CenterMask on unseen imagesCenterMask · OpenForensics detection (test-challenge)AP: 85.5 -> 0.03Hurt
use BlendMask for segmentation instead of detection-only boxesBlendMask · OpenForensics segmentation (test-dev)AP: 89.2Improved
use single-stage dense detectors instead of two-stage sparse onesforgery detectors · OpenForensics detection/segmentationoLRP FP and FNMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy