Merge branch 'abhisheks008:main' into main

Author: aaradhyasinghgaur

Age and Sex Prediction/Web App/README.md

@@ -0,0 +1,29 @@
+## Web Application for Age and Sex Prediction using Flask
+
+### Goal 🎯
+The main goal is to develop a user-friendly web application that predicts age and gender from human images using deep learning models. This application will allow users to upload images and receive predictions.
+
+### Model used for the Web App 🧮
+The backend of this web application utilizes Convolutional Neural Networks (CNNs) for age and gender prediction. It has been trained on [UTKFace Dataset](https://www.kaggle.com/datasets/jangedoo/utkface-new) dataset that contains over 20000 facial images for 30 epochs.
+
+Key Details:
+* **Model Architecture:** It utilizes Convolutional Neural Network.
+* **Data Processing**: Input images are converted to Grayscale images, resized to 128x128 pixels, and normalized.
+* **Frameworks and Libraries**: The model is implemented using TensorFlow and Keras for deep learning tasks, and OpenCV for image processing.
+* **Model Training**: The model uses categorical cross-entropy loss and Mean Average Error for gender and age respectively and the Adam optimizer. And sigmoid and ReLU activation functions are used for gender and age respectively. It has been trained for 30 epochs.
+* **Deployment**: The model is deployed using Flask, providing a web application to interact with the deep learning model efficiently.
+This setup enhances user interface allowing real-time predictions.
+
+
+### Video Demonstration 🎥
+
+[!()](https://github.com/theiturhs/DL-Simplified/assets/96874023/03bdb810-4b52-4606-9bc4-1384fcecb595)
+
+### Signature ✒️
+**Shruti Shrivastava**
+
+[![LinkedIn](https://img.shields.io/badge/LinkedIn-0077B5?style=for-the-badge&logo=linkedin&logoColor=white)](https://www.linkedin.com/in/shrutikshrivastava/)
+[![Gmail](https://img.shields.io/badge/Gmail-D14836?style=for-the-badge&logo=gmail&logoColor=white)](mailto:[email protected])
+[![Carrd](https://img.shields.io/badge/carrd-000000?style=for-the-badge&logo=carrd&logoColor=white)](https://theiturhs.carrd.co/)
+[![Kaggle](https://img.shields.io/badge/kaggle-0077B5?style=for-the-badge&logo=kaggle&logoColor=white)](https://www.kaggle.com/theiturhs)
+[![GitHub](https://img.shields.io/badge/github-000000?style=for-the-badge&logo=github&logoColor=white)](https://github.com/theiturhs)

Age and Sex Prediction/Web App/app.py

@@ -0,0 +1,101 @@
+from flask import Flask, render_template, request, redirect, url_for, session
+import os
+import numpy as np
+import cv2
+import tensorflow as tf
+import sys
+
+sys.stdout.reconfigure(encoding='utf-8')
+
+app = Flask(__name__)
+app.config['JSON_AS_ASCII'] = False
+app.secret_key = os.urandom(24)
+
+UPLOAD_FOLDER = 'static'
+app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
+
+# loading the model
+# model is saved in static folder
+model = tf.keras.models.load_model('static/age_and_gender_prediction_model.h5')
+
+# gender mapping
+# if gender prediction = 0, the predition is MALE
+# else FEMAL
+gender_dict = {0: 'Male', 1: 'Female'}
+
+# Function to extract face from uploaded image
+# The dataset on which model was trained contained only face images
+# This function finds the face area using
+# a haar cascade designed by OpenCV to detect the frontal face
+def extract_face():
+    
+    img = cv2.imread('static/uploaded_img.jpg') # reading uploaded image
+    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # converted to grayscale
+    
+    # OpenCV's CascadeClassifier to load a pre-trained Haar cascade for detecting frontal faces
+    haar_cascade = cv2.CascadeClassifier('static/haarcascade_frontalface_default.xml')
+    faces_rect = haar_cascade.detectMultiScale(gray_img, scaleFactor=1.1, minNeighbors=9)
+    extracted_faces = []
+
+    if len(faces_rect) == 0:
+        cv2.imwrite('static/extracted.jpg', img) # if no coordinates are detected, its only face image
+    else:
+        extracted_faces = []
+        for (x, y, w, h) in faces_rect:
+            face = img[y:y+h, x:x+w]
+            extracted_faces.append(face)
+        concatenated_faces = cv2.hconcat(extracted_faces)
+        cv2.imwrite('static/extracted.jpg', concatenated_faces) # face extracted image - saved as extracted.jpg
+
+    if len(faces_rect) != 0:
+        for (x, y, w, h) in faces_rect:
+            cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), thickness=2)
+        cv2.imwrite('static/uploaded_img.jpg', img) # original uploaded image where face is marked
+
+# Function to extract features from the image
+# This is the pre-process technique
+# which was applied on original dataset before training
+def extract_features(images):
+    img = cv2.imread(images, cv2.IMREAD_GRAYSCALE)  # Read image in grayscale
+    img = cv2.resize(img, (128, 128))  # Resize image
+    img = np.array(img)  # Convert image to numpy array
+    features = img.reshape(1, 128, 128, 1)  # Reshape to match input shape
+    return features
+
+# Function to predict gender and age
+def predict_result():
+    X = extract_features('static/extracted.jpg')
+    X = X/255.0
+    pred = model.predict(X.reshape(1, 128, 128, 1))
+    pred_gender = gender_dict[round(pred[0][0][0])]
+    pred_age = round(pred[1][0][0])
+    return pred_gender, pred_age
+
+@app.route('/prediction')
+def prediction():
+    image_filename = 'static/extracted.jpg'
+    gender, age = predict_result()
+    if image_filename:
+        return render_template('prediction_page.html', age=age, gender=gender)
+    else:
+        return render_template('not_found.html')  # Redirect to the 'not_found' page
+
+@app.route('/', methods=['GET', 'POST'])
+def home():
+    if request.method == 'POST' and 'face_image' in request.files:
+        face_image = request.files['face_image']
+        if face_image.filename != '':
+            image_filename = 'uploaded_img.jpg'
+            image_path = os.path.join(app.config['UPLOAD_FOLDER'], image_filename)
+            face_image.save(image_path)
+            session['image_filename'] = image_filename
+            extract_face()
+            return redirect(url_for('prediction'))  # Redirect to the 'prediction' route
+    return render_template('index.html')
+
+@app.route('/not_found')
+def not_found():
+    return render_template('not_found.html')
+
+if __name__ == "__main__":
+    app.run(debug=True, port=8000)