mp.py

from mtcnn.mtcnn import MTCNN
import cv2
from PIL import Image
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
from tensorflow.keras.utils import to_categorical
import numpy as np
import pandas as pd
import os

detector = MTCNN()

def extract_face(filename, required_size=(160, 160)):
    # Load the image
    image = cv2.imread(filename)
    if image is None:
        print(f"Could not load image at {image_path}")
        return None
    # Convert the image to RGB (MTCNN expects RGB)
    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    # Detect faces in the image
    results = detector.detect_faces(image)
    # Extract the bounding box from the first face
    x1, y1, width, height = results[0]['box']
    x2, y2 = x1 + width, y1 + height
    # Extract the face
    face = image[y1:y2, x1:x2]
    # Resize pixels to the model size
    image = Image.fromarray(face)
    image = image.resize(required_size)
    face_array = np.asarray(image)
    return face_array

csv_path = 'train_small.csv'
data = pd.read_csv(csv_path)

# Assuming your images are stored in a directory
image_dir = 'train_small'

# Initialize lists to hold processed images and labels
processed_images = []
labels = []

for _, row in data.iterrows():
    image_path = os.path.join(image_dir, row['File Name'])
    print(image_path)
    face = extract_face(image_path)  # Make sure this function returns an appropriately resized image
    if face is not None:  # Ensure a face was detected
        processed_images.append(face)
        labels.append(row['Category'])

# Convert the processed images and labels into numpy arrays
processed_images = np.array(processed_images)
labels = np.array(labels)

# Encode the labels
label_encoder = LabelEncoder()
encoded_labels = label_encoder.fit_transform(labels)
encoded_labels = to_categorical(encoded_labels)

# Split the dataset
X_train, X_test, y_train, y_test = train_test_split(processed_images, encoded_labels, test_size=0.2, random_state=42)