How tenka04/antifake-video-detection works
AntiFake is a video-authenticity analysis product with a React dashboard, a FastAPI upload/URL-analysis API, persistent scan history, and a model inference runtime. The main workflow accepts a video in the Analyze page, sends it to the API, routes analysis to inference, and returns a verdict and detector results for display; dashboard and history pages fetch persisted data. The README describes a seven-detector ensemble, while the sampled runtime excerpt shows detector scoring and fusion but not every detector implementation. The graph preserves documented capabilities and marks relationships supported only by documentation accordingly.
Main entry points: api.ts API client · Analyze.tsx Analyze page · App.tsx App routes · main.py FastAPI routes
Components 21
Web client
- App routes
frontend/src/App.tsx - Analyze page
frontend/src/pages/Analyze.tsx - Video inputs
frontend/src/components/video/VideoDropzone.tsx - Video preview
frontend/src/components/video/VideoPreview.tsx - API client
frontend/src/services/api.ts - Analysis results
frontend/src/components/analysis/ResultCard.tsx - Dashboard
frontend/src/pages/Dashboard.tsx - Scan history
frontend/src/pages/History.tsx - Theme settings
frontend/src/pages/Settings.tsx
API and history
- FastAPI routes
backend/app/main.py - SQLite history
backend/app/database.py - Inference client
backend/app/model_client.py
Video inference
- gRPC service
model/server.py - Forensic detectors
poc_runtime.py - Score fusion
poc_runtime.py - VideoMAE detector
backend/app/models/videomae_detector.py - Video inspection
backend/app/preprocessing/video_loader.py - Frame sampling
backend/app/preprocessing/frame_sampler.py
Standalone analysis
- Gradio interface
app.py - HTTP model server
model/http_server.py
Other components
- Video analyst
Connections 24
- Video analyst → Analyze page: submits video
README.md - Analyze page → Video inputs: collects input
frontend/src/pages/Analyze.tsx - Analyze page → Video preview: previews video
frontend/src/pages/Analyze.tsx - Analyze page → API client: requests analysis
frontend/src/pages/Analyze.tsx - API client → FastAPI routes: sends REST request
README.md - FastAPI routes → Inference client: routes analysis
backend/app/main.py - FastAPI routes → SQLite history: records and queries
backend/app/main.py - Inference client → gRPC service: calls gRPC
backend/app/model_client.py - gRPC service → Forensic detectors: runs analysis
model/server.py - Forensic detectors → Score fusion: fuses scores
poc_runtime.py - Score fusion → VideoMAE detector: includes visual model
README.md - VideoMAE detector → Video inspection: uses preprocessing
backend/app/models/videomae_detector.py - Frame sampling → Video inspection: reads video
backend/app/preprocessing/frame_sampler.py - FastAPI routes → API client: returns result
backend/app/main.py - API client → Analysis results: provides results
frontend/src/pages/Analyze.tsx - App routes → Analyze page: routes to
frontend/src/App.tsx - App routes → Dashboard: routes to
frontend/src/App.tsx - App routes → Scan history: routes to
frontend/src/App.tsx - App routes → Theme settings: routes to
frontend/src/App.tsx - Dashboard → API client: fetches statistics
frontend/src/pages/Dashboard.tsx - Scan history → API client: fetches history
frontend/src/pages/History.tsx - Video analyst → Gradio interface: runs standalone scan
README.md - Gradio interface → Forensic detectors: uses runtime
app.py - HTTP model server → Forensic detectors: serves analysis
model/http_server.py