🔬 FORENSIQ
Physics-Based, Multi-Agent Forensic Framework for Explainable Deepfake Detection
7 Independent Forensic Agents • Bayesian Evidence Synthesis • Court-Admissible Reports
Supported: JPEG, PNG, WebP, BMP, TIFF
Agents: Optical Physics • Sensor • Generative Model • Statistical • Semantic • Metadata • Text
Engine: Bayesian Evidence Synthesis with Independence Correction
Agents: Optical Physics • Sensor • Generative Model • Statistical • Semantic • Metadata • Text
Engine: Bayesian Evidence Synthesis with Independence Correction
⚡ VLM Status: Semantic & Text agents require HF Inference credits.
Without credits, 5/7 agents run (signal processing only). For full 7/7 analysis
including anatomy, lighting, and physics checks, ensure your HF account has
active inference credits
or a PRO subscription.
Upload an image and click Analyze
EXIF Metadata
Export Forensic Report
Export the complete analysis in your preferred format. Reports are professionally formatted using Qwen2.5-72B-Instruct when available.
FORENSIQ v1.0 — Physics-Based Multi-Agent Forensic Framework
7 Agents • 127+ Physical Constraints • Bayesian Evidence Synthesis • Court-Admissible Reports
Powered by Qwen2.5-VL for semantic analysis • Signal processing via NumPy/SciPy
7 Agents • 127+ Physical Constraints • Bayesian Evidence Synthesis • Court-Admissible Reports
Powered by Qwen2.5-VL for semantic analysis • Signal processing via NumPy/SciPy