How nyn-05/major_project works
Deepfake Video Detection for KYC takes an uploaded video through frame sampling and quality filtering, rPPG and visual feature extraction, then fused quantum-classical classification to produce a REAL/FAKE verdict. The repository also includes a React UI and a standard-library API server. The README documents the main stage handoffs; some internal edges are based on SOURCE INDEX references, and the frontend-to-backend/runtime links are not confirmed by the sampled excerpts.
Main entry points: quantum/pipeline.py Quantum Pipeline · run_pipeline.py Video Orchestrator · RPPG/pipeline.py rPPG Pipeline · frame/pipeline.py Frame Sampling
Components 20
Web Experience
- Review UI
frontend/src/App.jsx - HTTP API
frontend/src/api.js - API Server
frontend/server.py
Frame Intake
- Frame Sampling
WORKING/frame/pipeline.py - YOLO Face Detection
WORKING/frame/detector.py - Frame Quality
WORKING/frame/processing.py
Evidence Features
- rPPG Pipeline
WORKING/RPPG/pipeline.py - Face ROI Tracking
WORKING/RPPG/face_roi.py - Pulse Reconstruction
WORKING/RPPG/signal_extraction.py - Physiological Features
WORKING/RPPG/features.py - Visual Feature Pipeline
WORKING/visual/pipeline.py - Visual Extractor
WORKING/visual/extractor.py
Quantum Decision
- Fused Data Builder
WORKING/quantum/data.py - QAOA Selection
WORKING/quantum/qaoa.py - Hybrid VQC
WORKING/quantum/vqc.py - Quantum Pipeline
WORKING/quantum/pipeline.py - Model Evaluation
WORKING/quantum/evaluation.py - Verdict Explanation
WORKING/quantum/explain.py
Other components
- KYC Operator
- Video Orchestrator
WORKING/run_pipeline.py
Connections 21
- KYC Operator → Review UI: uploads video
- Review UI → HTTP API: requests analysis
frontend/src/App.jsx - HTTP API → API Server: calls API
- API Server → Video Orchestrator: invokes pipeline
- Video Orchestrator → Frame Sampling: samples frames
WORKING/run_pipeline.py - Frame Sampling → YOLO Face Detection: detects faces
WORKING/frame/pipeline.py - Frame Sampling → Frame Quality: filters quality
WORKING/frame/pipeline.py - Video Orchestrator → rPPG Pipeline: extracts rPPG
WORKING/run_pipeline.py - rPPG Pipeline → Face ROI Tracking: extracts ROIs
WORKING/RPPG/pipeline.py - rPPG Pipeline → Pulse Reconstruction: reconstructs pulse
WORKING/RPPG/pipeline.py - rPPG Pipeline → Physiological Features: computes features
WORKING/RPPG/pipeline.py - Video Orchestrator → Visual Feature Pipeline: extracts visual features
- Visual Feature Pipeline → Visual Extractor: extracts crops
WORKING/visual/pipeline.py - Video Orchestrator → Quantum Pipeline: runs fused decision
WORKING/run_pipeline.py - Visual Feature Pipeline → Fused Data Builder: builds fused set
WORKING/visual/pipeline.py - Quantum Pipeline → Fused Data Builder: loads fused data
WORKING/quantum/pipeline.py - Quantum Pipeline → QAOA Selection: selects features
WORKING/quantum/pipeline.py - Quantum Pipeline → Hybrid VQC: scores features
WORKING/quantum/pipeline.py - Quantum Pipeline → Model Evaluation: evaluates model
WORKING/quantum/pipeline.py - Quantum Pipeline → Verdict Explanation: builds verdict
WORKING/quantum/pipeline.py - Quantum Pipeline → Review UI: returns verdict