How vikramforsk2019/vehicle-plate-identify works
Project type and purpose
• This is a full-stack, single-service web application.
• Purpose: users upload car images via a browser UI; the server runs YOLOv3 to detect license plates, crops them, runs OCR, and returns an annotated image or plate text.
File structure
Top-level files:
• Procfile – deploy on Heroku or similar.
• app.py – Flask entry point that wires routes to detection logic.
• requirements.txt – Python dependencies (Flask, TensorFlow/PyTorch, OpenCV, OCR libs).
• darknet-yolov3.cfg, classes.names, lapi.weights – YOLOv3 model definition, labels, and trained weights.
• ocr_core.py – logic to perform OCR on cropped plate images.
Directories:
• static/ – client assets (index.js, style.css), upload folder for images, detect_cut folder for cropped plates.
• templates/ – HTML templates for upload form and results.
README for additional insights
• Detailed pipeline for data preparation, training Darknet YOLOv3 on plate datasets.
• Instructions for using the Flask API.
• References to Google Drive folder structure (training artifacts).
Detail
• Every folder that holds runtime artifacts (static/uploads, static/detect_cut).
• Call out config & weight files loading by the YOLO component.
• Flow from cropping to OCR and back to the user.