How nimo94/smiloai works
SmiloAI is an offline-first dental diagnostic workstation. A practitioner uses the UI to submit RGB or X-ray scans, choose a single/batch, sequential, or custom-flow run, and review annotated findings and clinical reports. The README documents specialist models, routing, confidence/NMS filtering, and Auto Pilot training; the sampled main.py also shows UI serving, model unload/reload endpoints, and training events. The main implementation is concentrated in one large file, so several domain components share its path; other runtime internals were not sampled.
Main entry points: main.py API & Engine · index.html Workstation UI · AUTOPILOT_MODELS Routing Models · model Local Models
Components 17
Workstation UI
- Workstation UI
Sourcecodes/index.html
Diagnostic Runtime
- API & Engine
Sourcecodes/main.py - Model Discovery
Sourcecodes/main.py - Local Models
Sourcecodes/model - Scan Modes
Sourcecodes/main.py - Flow Builder
Sourcecodes/main.py - Specialist Inference
Sourcecodes/main.py - Smart Routing
Sourcecodes/main.py - Confidence & NMS
Sourcecodes/main.py
Auto Pilot Studio
- Routing Models
Sourcecodes/AUTOPILOT/AUTOPILOT_MODELS - Auto Pilot Training
Sourcecodes/main.py - Training Data
Sourcecodes/AUTOPILOT/Training_Data - Training Feedback
Sourcecodes/main.py
Clinical Output
- Visual Annotations
Sourcecodes/main.py - Clinical Reports
Sourcecodes/main.py
Other components
- Practitioner
- Streaming LLM
Connections 20
- Practitioner → Workstation UI: uses
README.md - Workstation UI → API & Engine: submits scans
README.md - API & Engine → Workstation UI: serves
Sourcecodes/main.py - API & Engine → Model Discovery: discovers models
README.md - Model Discovery → Local Models: loads models
README.md - API & Engine → Scan Modes: offers
README.md - API & Engine → Flow Builder: executes flows
README.md - Scan Modes → Specialist Inference: runs inference
README.md - Flow Builder → Specialist Inference: runs specialists
README.md - Scan Modes → Smart Routing: routes scans
README.md - Smart Routing → Routing Models: uses router
README.md - Specialist Inference → Confidence & NMS: filters predictions
README.md - Smart Routing → Confidence & NMS: filters predictions
Sourcecodes/main.py - Confidence & NMS → Visual Annotations: annotates findings
README.md - Visual Annotations → Clinical Reports: feeds reports
README.md - Workstation UI → Auto Pilot Training: starts training
README.md - Auto Pilot Training → Training Data: trains on
README.md - Auto Pilot Training → Training Feedback: publishes metrics
Sourcecodes/main.py - Auto Pilot Training → Routing Models: exports model
README.md - Workstation UI → Streaming LLM: requests assistance
README.md