How dhruvtalnewar01/atmanirbhar-ai works
AtmaNirbhar AI is a road-hazard perception and driver-alert application: a user submits an image or video, the backend analyzes frames, and the frontend presents detections, spatial context, and alerts. The backend entry point owns pipeline orchestration; the frontend page owns API submission and visualization. The README documents additional audio, depth, collision, and replanning capabilities, but sampled excerpts do not establish every internal call, so those uncertain links are omitted.
Main entry points: main.py Analysis API · page.tsx Inspection UI · api.ts API Client · schemas.py Response Models
Components 20
Driver Interface
- Inspection UI
frontend/src/app/page.tsx - API Client
frontend/src/lib/api.ts - Canvas Overlays
frontend/src/lib/canvas-renderer.ts - 3D Scene Viewer
frontend/src/components/viewer/LidarVisualizer.tsx - Alert Audio
frontend/src/lib/audio-alert.ts
Analysis Service
- Analysis API
backend/app/main.py - Response Models
backend/app/schemas.py
Perception Pipeline
- Frame Enhancement
backend/app/pipeline/enhance.py - Object Detection
backend/app/pipeline/detect.py - Object Tracking
backend/app/pipeline/track.py - VLM Verification
backend/app/pipeline/verify.py - Siren Analysis
backend/app/pipeline/audio.py
Spatial Safety
- Monocular Distance
backend/app/pipeline/distance.py - Ego Motion
backend/app/pipeline/motion.py - Collision Warnings
backend/app/pipeline/collision.py - Risk Scoring
backend/app/pipeline/risk.py - Pipeline Timing
backend/app/pipeline/timing.py - Voice Alerts
backend/app/pipeline/voice.py
Other components
- Driver / User
- Roboflow Inference
Connections 20
- Driver / User → Inspection UI: submits media
- Inspection UI → API Client: calls analysis
- API Client → Analysis API: uploads media
- Analysis API → Frame Enhancement: enhances frames
- Analysis API → Object Detection: detects objects
- Analysis API → Object Tracking: tracks detections
- Analysis API → VLM Verification: verifies tracks
- Analysis API → Siren Analysis: analyzes sirens
- Analysis API → Monocular Distance: estimates distance
- Analysis API → Ego Motion: estimates motion
- Analysis API → Collision Warnings: checks collisions
- Analysis API → Risk Scoring: scores risk
- Analysis API → Pipeline Timing: records timing
- Analysis API → Voice Alerts: requests alerts
- Object Detection → Roboflow Inference: requests inference
- Analysis API → Response Models: builds response
- Response Models → API Client: returns results
- Inspection UI → Canvas Overlays: renders frames
- Inspection UI → 3D Scene Viewer: shows spatial scene
- Inspection UI → Alert Audio: plays alerts