01 Session Analysis
// Detection distribution — this session
No videos scanned yet.
▶ Run your first scan
// Model accuracy — test set
98.2%
Detection Accuracy
CrossViT
Architecture
224px
Input Size
30
Max Frames
Haar
Face Detector
02 System Status
// Runtime Log
Flask dev server started on http://127.0.0.1:5000 ONLINE
CrossEfficientViT loaded from model/best_cross_vit.pth READY
Haar cascade initialized — haarcascade_frontalface_default.xml READY
Class mapping loaded — REAL / FAKE from model/classes.json OK
Session stats — Total: 0 | Real: 0 | Fake: 0 CLEAN
// LIVE SCAN HISTORY
STREAM OFFLINE ▷ OPEN LIVE PAGE
CURRENT VERDICT
WAITING
CONFIDENCE
-
FRAMES ANALYZED
0
STREAM FPS
0
TIMESESSIONVERDICTCONF %
No live scans yet - start a live session
Ready to Analyse?
Upload a video or start a live stream for real-time deepfake detection.
● GO LIVE ▶ UPLOAD VIDEO