How mariammm-bit/multi-crises-detection works
This is a small smart-building fire, smoke, and occupant monitoring application. A user selects image or camera mode; main.py obtains frames, runs fire/smoke and person detection, calculates a risk level, annotates the frame, and displays it. Separate scripts provide person-only image detection and YOLO training. The README gives no further product detail; model internals and dataset contents are not shown.
Main entry points: main.py Frame processing · human_detection.py Person image detection · danger_analyzer.py Risk analysis · train.py YOLO training
Components 15
Live monitoring
- Mode selection
main.py - Image input
main.py - Camera input
main.py - Frame processing
main.py - Fire/smoke model
- Person model
- Risk analysis
danger_analyzer.py - Risk panel
danger_analyzer.py - Monitor display
main.py
Standalone detection
- Person image detection
human_detection.py - Person model
- Image result
human_detection.py
Model training
- YOLO training
train.py - Dataset config
data.yaml
Other components
- Operator
Connections 14
- Operator → Mode selection: selects mode
main.py - Mode selection → Image input: starts image mode
main.py - Mode selection → Camera input: starts camera mode
main.py - Image input → Frame processing: passes image
main.py - Camera input → Frame processing: passes frames
main.py - Frame processing → Fire/smoke model: runs inference
main.py - Frame processing → Person model: runs inference
main.py - Frame processing → Risk analysis: submits counts
main.py - Frame processing → Risk panel: requests annotation
main.py - Risk analysis → Risk panel: provides risk level
main.py - Frame processing → Monitor display: returns annotated frame
main.py - Person image detection → Person model: runs inference
human_detection.py - Person image detection → Image result: saves and displays
human_detection.py - YOLO training → Dataset config: loads configuration
train.py