cnn_updated_code

import os
import pandas as pd
import numpy as np
import cv2
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from sklearn.model_selection import train_test_split
from tensorflow.keras.optimizers import Adam, SGD, RMSprop


# Define paths
dataset_csv = "dataset_new.csv"
image_folder = "E:/image labelling/annotated_images"  # Update path accordingly

# Load dataset
df = pd.read_csv(dataset_csv)

# Parameters
IMG_SIZE = 64  # Resize all images to 64x64
NUM_CLASSES = 13  

# Load images and labels
X = []
y = []

for index, row in df.iterrows():
    image_path = os.path.join(image_folder, row["Image_Name"])
    
    # Read and preprocess image
    if os.path.exists(image_path):
        image = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)  # Load as grayscale
        image = cv2.resize(image, (IMG_SIZE, IMG_SIZE))  # Resize
        image = image / 255.0  # Normalize pixel values (0-1)
        X.append(image)
        y.append(row["Label"])

# Convert to NumPy arrays
X = np.array(X).reshape(-1, IMG_SIZE, IMG_SIZE, 1)  # Add channel dimension
y = np.array(y)

# Train-test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# one-hot encoding
y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)
y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)

# CNN model
model = keras.Sequential([
    layers.Conv2D(32, (3, 3), activation="relu", input_shape=(IMG_SIZE, IMG_SIZE, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation="relu"),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(128, (3, 3), activation="relu"),
    layers.MaxPooling2D((2, 2)),
    layers.Flatten(),
    layers.Dense(128, activation="relu"),
    layers.Dense(NUM_CLASSES, activation="softmax")
])

# Compile model
model.compile(optimizer=Adam(learning_rate=0.001), loss="categorical_crossentropy", metrics=["accuracy"])

# Train model
history = model.fit(X_train, y_train, epochs=20, batch_size=16, validation_data=(X_test, y_test))

# Evaluate model
test_loss, test_acc = model.evaluate(X_test, y_test)
print(f"Test accuracy: {test_acc:.4f}")

# Save the trained model
model.save("cnn_image_classifier.h5")
print("Model saved as 'cnn_image_classifier.h5'")


# test on an image sample
image = cv2.imread("E:/image labelling/reference_images/apporva.JPG", cv2.IMREAD_GRAYSCALE)
image = cv2.resize(image, (64, 64))  # Resize to match model input size
image = image / 255.0  # Normalizatiobn

# Reshape to match CNN input (batch_size, height, width, channels)
image = np.reshape(image, (1, 64, 64, 1)).astype(np.float32)

# Predict
prediction = model.predict(image)
predicted_label = np.argmax(prediction)

# Print results
print(f"Predicted Label: {predicted_label}")
print(f"Prediction Probabilities: {prediction[0]}")