Changelog

Updates and improvements to Resemble AI
Back to changelog
July 1, 2026

Visual DFD model update — V-JEPA 2 video backbone

Deployed the 20260630 visual detection release. The video model backbone moved to V-JEPA 2 (F1 0.985→0.990, AUC 0.993→0.997), image detection improved on passport/document fraud (average accuracy 84.4%→92.9%), and the report heatmap and timeline tree-view were refreshed with edge-highlighting and representative-frame analysis.