How nawaf-rayhan585/lpr-vehicle-license-detection works
LPR is a Streamlit app that accepts a traffic video, detects vehicles, reads and deduplicates plate text, then presents results and downloadable outputs. The principal workflow is implemented in one file, so several nodes represent distinct responsibilities within main.py rather than separate modules. The README documents the product capabilities; runtime relationships shown here are supported by the excerpted source.
Main entry points: main.py Video processor · yolov8n.pt YOLO weights
Components 15
User interface
- Streamlit interface
main.py - Input staging
main.py - Download and preview
main.py
Recognition pipeline
- Model loader
main.py - YOLO weights
yolov8n.pt - Video processor
main.py - Vehicle detection
main.py - Vehicle class filter
main.py - Vehicle crop
main.py - EasyOCR reader
main.py
Results and outputs
- Plate validation and dedupe
main.py - Frame annotation
main.py - Processed video
main.py - Plate report
main.py
Other components
- Video uploader
Connections 16
- Video uploader → Streamlit interface: uploads video
main.py - Streamlit interface → Input staging: saves upload
main.py - Streamlit interface → Model loader: loads models
main.py - Model loader → YOLO weights: loads weights
main.py - Streamlit interface → Video processor: runs processing
main.py - Video processor → Vehicle detection: detects vehicles
main.py - Video processor → Vehicle class filter: filters classes
main.py - Video processor → Vehicle crop: crops vehicles
main.py - Video processor → EasyOCR reader: calls run_ocr
main.py - EasyOCR reader → Plate validation and dedupe: yields text
main.py - Video processor → Frame annotation: draws boxes and text
main.py - Video processor → Processed video: writes frames
main.py - Video processor → Plate report: returns unique plates
main.py - Streamlit interface → Download and preview: offers results
main.py - Download and preview → Plate report: downloads JSON
main.py - Download and preview → Processed video: previews and downloads
main.py