yolo_module/ocr_reader.py

"""
License plate OCR using EasyOCR.
Accepts a cropped plate image (numpy array) and returns extracted text.
"""

import re
import numpy as np
import easyocr

# Singleton reader to avoid reloading the model on every call
_reader = None


def get_reader() -> easyocr.Reader:
    global _reader
    if _reader is None:
        _reader = easyocr.Reader(["en"], gpu=False, verbose=False)
        print("[OCR] EasyOCR reader initialised")
    return _reader


def read_plate(crop: np.ndarray) -> str:
    """
    Extract license plate text from a BGR/RGB crop.

    Returns:
        Cleaned plate string (e.g. "KA05MN1234") or "" if nothing detected.
    """
    if crop is None or crop.size == 0:
        return ""

    reader = get_reader()
    results = reader.readtext(crop, detail=1, paragraph=False)

    if not results:
        return ""

    # Pick the result with highest confidence
    best = max(results, key=lambda r: r[2])
    raw_text = best[1]

    return _clean_plate(raw_text)


def _clean_plate(text: str) -> str:
    """
    Normalise OCR output to uppercase alphanumeric only.
    Common OCR confusions corrected: O↔0, I↔1, S↔5.
    """
    text = text.upper().strip()
    text = re.sub(r"[^A-Z0-9]", "", text)
    return text