Merge pull request #941 from Dharun235/mask-detection-project

Created the folder containing files for mask detection

Image processing/Mask-detection/MaskDetect.py

@@ -0,0 +1,51 @@
+import cv2
+import tensorflow as tf
+import numpy as np
+
+# Load a pre-trained MobileNetV2 mask detection model (change to your model path or URL if available)
+model = tf.keras.models.load_model('mask_detector_mobilenetv2.h5')  # Make sure the model is in the same folder or provide path
+
+# Function to preprocess the image for MobileNetV2
+def preprocess_image(face):
+    face_resized = cv2.resize(face, (224, 224))  # Resize to MobileNetV2 input size
+    face_normalized = face_resized / 255.0       # Normalize pixel values
+    face_expanded = np.expand_dims(face_normalized, axis=0)  # Add batch dimension
+    return face_expanded
+
+# Function to perform mask detection
+def detect_mask(frame):
+    # Convert the frame to a tensor and preprocess it
+    face_preprocessed = preprocess_image(frame)
+    prediction = model.predict(face_preprocessed)
+
+    # MobileNetV2 model output for binary classification (Mask / No Mask)
+    mask_probability = prediction[0][0]  # Get the single output probability for "with_mask"
+
+    # Set the threshold for mask detection
+    threshold = 0.5
+    label = "Mask" if mask_probability > threshold else "No Mask"
+    color = (0, 255, 0) if label == "Mask" else (0, 0, 255)
+
+    return label, color
+
+# Initialize video capture
+cap = cv2.VideoCapture(0)
+
+while True:
+    ret, frame = cap.read()
+    if not ret:
+        break
+
+    # Pre-process each frame for mask detection
+    label, color = detect_mask(frame)
+
+    # Draw label on the frame
+    cv2.putText(frame, f'{label}', (30, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, color, 2)
+    cv2.imshow("Mask Detection", frame)
+
+    # Exit on 'q' key press
+    if cv2.waitKey(1) & 0xFF == ord('q'):
+        break
+
+cap.release()
+cv2.destroyAllWindows()

Image processing/Mask-detection/Model-training.py

@@ -0,0 +1,119 @@
+import tensorflow as tf
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
+from tensorflow.keras.applications import MobileNetV2
+from tensorflow.keras.layers import Dense, GlobalAveragePooling2D
+from tensorflow.keras.models import Model
+from tensorflow.keras.optimizers import Adam
+import os
+import cv2
+import xml.etree.ElementTree as ET
+import numpy as np
+
+# Function to parse XML annotations
+def parse_annotations(annotations_path):
+    data = []
+
+    for xml_file in os.listdir(annotations_path):
+        if xml_file.endswith('.xml'):
+            tree = ET.parse(os.path.join(annotations_path, xml_file))
+            root = tree.getroot()
+            image_file = root.find('filename').text  # Get image filename without path
+
+            # For each object in the annotation
+            objects = root.findall('object')
+            for obj in objects:
+                label = obj.find('name').text  # Get label
+                # Map labels to binary (1 for with_mask, 0 for without_mask)
+                data.append((image_file, 1 if label == 'with_mask' else 0))
+
+    return data
+
+# Paths
+annotations_path = 'dataset/annotations'
+images_path = 'dataset/images'
+
+# Parse annotations to get image paths and labels
+data = parse_annotations(annotations_path)
+
+# Create a function to load images and their labels
+def load_image_and_label(image_name):
+    image_path = os.path.join(images_path, image_name)
+
+    # Check if the image exists
+    if not os.path.exists(image_path):
+        print(f"Image {image_name} not found.")
+        return None, None
+
+    image = cv2.imread(image_path)
+    image = cv2.resize(image, (224, 224))  # Resize to MobileNetV2 input size
+    image = image / 255.0  # Normalize image
+    return image
+
+# Prepare the dataset
+images = []
+labels = []
+
+for image_name, label in data:
+    img = load_image_and_label(image_name)
+    if img is not None:  # Only add valid images
+        images.append(img)
+        labels.append(label)
+
+images = np.array(images)
+labels = np.array(labels)
+
+# Create a training-validation split
+split_index = int(0.8 * len(images))
+train_images = images[:split_index]
+train_labels = labels[:split_index]
+val_images = images[split_index:]
+val_labels = labels[split_index:]
+
+# Create ImageDataGenerators
+train_datagen = ImageDataGenerator(
+    rotation_range=20,
+    zoom_range=0.2,
+    width_shift_range=0.2,
+    height_shift_range=0.2,
+    shear_range=0.2,
+    horizontal_flip=True,
+    fill_mode="nearest"
+)
+
+val_datagen = ImageDataGenerator()
+
+# Create generators
+train_generator = train_datagen.flow(train_images, train_labels, batch_size=32)
+val_generator = val_datagen.flow(val_images, val_labels, batch_size=32)
+
+# Load MobileNetV2
+base_model = MobileNetV2(weights="imagenet", include_top=False, input_shape=(224, 224, 3))
+
+# Freeze base model layers
+for layer in base_model.layers:
+    layer.trainable = False
+
+# Add custom top layers
+x = base_model.output
+x = GlobalAveragePooling2D()(x)
+predictions = Dense(1, activation="sigmoid")(x)  # Sigmoid for binary classification
+
+# Create model
+model = Model(inputs=base_model.input, outputs=predictions)
+
+# Compile model
+model.compile(optimizer=Adam(learning_rate=1e-4), loss="binary_crossentropy", metrics=["accuracy"])
+
+# Train model
+epochs = 10
+history = model.fit(
+    train_generator,
+    steps_per_epoch=len(train_images) // 32,
+    validation_data=val_generator,
+    validation_steps=len(val_images) // 32,
+    epochs=epochs
+)
+
+# Save model
+model.save("mask_detector_mobilenetv2.h5")
+print("Model saved as mask_detector_mobilenetv2.h5")

Image processing/Mask-detection/README.md

@@ -0,0 +1,112 @@
+# Mask Detection Using MobileNetV2
+
+## Project Overview
+This project implements a real-time mask detection system using deep learning techniques. The model is based on the MobileNetV2 architecture, which is pre-trained on the ImageNet dataset. The system can classify individuals as wearing a mask or not wearing a mask using live video feeds from a webcam.
+
+## Table of Contents
+- [Features](#features)
+- [Requirements](#requirements)
+- [Dataset Structure](#dataset-structure)
+- [Installation](#installation)
+- [Training the Model](#training-the-model)
+- [Using the Model](#using-the-model)
+
+## Features
+- Real-time mask detection from webcam feed.
+- Utilizes MobileNetV2 for efficient image classification.
+- Simple and user-friendly interface.
+- Easy to modify and extend for other use cases.
+
+## Requirements
+- Python 3.6 or higher
+- TensorFlow
+- OpenCV
+- NumPy
+- Other standard libraries (included in the Python standard library)
+
+You can install the required libraries using pip:
+
+```bash
+pip install tensorflow opencv-python numpy
+```
+
+## Dataset Structure
+The dataset should contain two main folders: `annotations` and `images`. The structure should look like this:
+Dataset can be downloaded using the link: https://www.kaggle.com/datasets/andrewmvd/face-mask-detection/data 
+```
+dataset/
+│
+├── annotations/
+│   ├── example1.xml
+│   ├── example2.xml
+│   └── ...
+│
+└── images/
+    ├── example1.png
+    ├── example2.png
+    └── ...
+```
+
+### XML Annotation Format
+Each XML file corresponds to an image and should contain bounding box information and labels. Example XML structure:
+
+```xml
+<annotation>
+    <folder>images</folder>
+    <filename>example1.png</filename>
+    <size>
+        <width>512</width>
+        <height>366</height>
+        <depth>3</depth>
+    </size>
+    <object>
+        <name>with_mask</name>
+        <bndbox>
+            <xmin>100</xmin>
+            <ymin>150</ymin>
+            <xmax>200</xmax>
+            <ymax>250</ymax>
+        </bndbox>
+    </object>
+    <object>
+        <name>without_mask</name>
+        <bndbox>
+            <xmin>250</xmin>
+            <ymin>150</ymin>
+            <xmax>350</xmax>
+            <ymax>250</ymax>
+        </bndbox>
+    </object>
+</annotation>
+```
+
+## Installation
+1. Clone this repository to your local machine:
+   ```bash
+   git clone <repository_url>
+   cd <repository_name>
+   ```
+
+2. Install the required libraries as mentioned above.
+
+## Training the Model
+To train the mask detection model, follow these steps:
+
+1. Prepare your dataset or use the dataset given in this folder.
+2. Use the provided training script (`Model-training.py`) to train the model:
+   ```bash
+   python Model-training.py
+   ```
+
+3. After training, the model will be saved as `mask_detector_mobilenetv2.h5`.
+
+## Using the Model
+To use the trained model for real-time mask detection:
+
+1. Run the mask detection script (`MaskDetect.py`):
+   ```bash
+   python MaskDetect.py
+   ```
+
+2. The webcam feed will open, and the model will classify whether individuals are wearing a mask or not in real-time. Press 'q' to exit.
+

Project-Structure.md

@@ -622,6 +622,9 @@
   * Lane Line Detection [Open Cv]
     * [Gui](Image%20processing/Lane%20Line%20Detection%20[OPEN%20CV]/gui.py)
     * [Main](Image%20processing/Lane%20Line%20Detection%20[OPEN%20CV]/main.py)
+  * Mask-Detection
+    * [Maskdetect](Image%20processing/Mask-detection/MaskDetect.py)
+    * [Model-Training](Image%20processing/Mask-detection/Model-training.py)
   * Number Plate Detection
     * [Number Plate](Image%20processing/Number%20plate%20detection/number_plate.py)
     * Ocr-Notebook