NPR using CNN Keras convo2D and Max pooling.py

import cv2
import imutils
import pytesseract
import serial
import time
import os
import numpy as np
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense

pytesseract.pytesseract.tesseract_cmd = 'C:\Program Files\Tesseract-OCR\\tesseract'

# Load images from directory
folder_path = 'C:/img_dataset'
images = []

for file_name in os.listdir(folder_path):
    if file_name.endswith('.jpg') or file_name.endswith('.jpeg') or file_name.endswith('.png'):
        image_path = os.path.join(folder_path, file_name)
        image = cv2.imread(image_path)
        images.append(image)

# Define CNN model
model = Sequential()
model.add(Conv2D(32, (3, 3), input_shape=(100, 100, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(256, activation='sigmoid'))
model.add(Dense(128, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

# Train the model
X_train = []
y_train = []
for image in images:
    resized_image = cv2.resize(image, (100, 100))
    X_train.append(resized_image)
    y_train.append(1)  # assuming all images contain license plates

X_train = np.array(X_train)
y_train = np.array(y_train)
model.fit(X_train, y_train, epochs=10)

# Use the model to detect license plates
cap = cv2.VideoCapture(0)
number_plate_dataset = {
    "BIK 1234": "COM8",
    "ATM 2341": "COM8",
    "FFT 4567": "COM8"
}

while True:
    ret, frame = cap.read()
    frame = imutils.resize(frame, width=600)
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    gray = cv2.bilateralFilter(gray, 11, 17, 17)
    edged = cv2.Canny(gray, 30, 200)
    cnts, hierarchy = cv2.findContours(edged.copy(), cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
    image1 = frame.copy()
    cv2.drawContours(image1, cnts, -1, (0, 255, 0), 3)
    cv2.imshow("Contours", image1)
    cnts = sorted(cnts, key=cv2.contourArea, reverse=True)[:30]
    screenCnt = None
    for c in cnts:
        perimeter = cv2.arcLength(c, True)
        approx = cv2.approxPolyDP(c, 0.018 * perimeter, True)
        if len(approx) == 4:
            screenCnt = approx
            x, y, w, h = cv2.boundingRect(c)
            new_img = frame[y:y + h, x:x + w]
            cv2.imshow("Cropped Image", new_img)
            break
    if screenCnt is not None:
        cv2.drawContours(frame, [screenCnt], -1, (0, 255, 0), 3)
        cropped = cv2.cvtColor(new_img, cv2.COLOR_BGR2GRAY)
        cropped = cv2.threshold(cropped, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
        plate = pytesseract.image_to_string(cropped, lang='eng')
        print("Number plate is:", plate)
        if plate.strip() in number_plate_dataset:
            SerialObj = serial.Serial('COM8')
            SerialObj.baudrate = 9600
            SerialObj.bytesize = 8
            SerialObj.parity = 'N'
            SerialObj.stopbits = 1
            time.sleep(3)
            SerialObj.write(plate.encode())
            SerialObj.close()
            exit()
    cv2.imshow("Original Image", frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()