Merge pull request #1789 from UppuluriKalyani/master

Face_Orientation_Predictor

Awesome-face-operations/Face_Orientation_Predictor/README.md

@@ -0,0 +1,126 @@
+Sure! Below is a sample description for your README file on GitHub, detailing the task and providing instructions on how to run the code.
+
+---
+
+# Neural Network for Face Orientation Prediction
+
+This repository contains a neural network implementation for predicting the orientation of faces in grayscale images. The model is trained on a dataset of face images, and it can predict four possible orientations: straight, up, left, and right.
+
+## Project Overview
+
+This project involves the following steps:
+
+1. **Data Collection**: Download face images from a specified URL and save them to a local directory.
+2. **Preprocessing**: Resize the images to 32x30 pixels and normalize the pixel values to the range [0,1].
+3. **Model Training**: Train a neural network to classify the orientation of the faces.
+4. **Model Saving**: Save the trained model to a file for future use.
+5. **Model Loading and Prediction**: Load the saved model and use it to predict the orientation of new face images.
+6. **Visualization**: Display the images along with their predicted labels.
+
+## Getting Started
+
+### Prerequisites
+
+- Python 3.x
+- Required Python libraries: `numpy`, `requests`, `beautifulsoup4`, `opencv-python`, `matplotlib`, `Pillow`
+
+Install the required libraries using pip:
+
+```bash
+pip install numpy requests beautifulsoup4 opencv-python matplotlib Pillow
+```
+
+### Data Collection
+
+The script downloads face images from the following URL: [Carnegie Mellon University Face Images](https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/).
+
+```python
+# URL of the webpage
+main_url = "https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/"
+
+# Directory to save downloaded files
+base_directory = "/content/faces"
+
+# Downloading function
+# [Include the code for downloading files here]
+```
+
+### Preprocessing
+
+Resize images to 32x30 pixels and normalize them to the range [0,1].
+
+```python
+def resize_image(image_path):
+    img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
+    resized_img = cv2.resize(img, (32, 30))
+    return resized_img / 255.0
+```
+
+### Model Training
+
+Train the neural network on the preprocessed images and save the model.
+
+```python
+# Define and train the neural network
+input_size = 960  # 32 * 30
+hidden_size = 100
+output_size = 4
+
+nn = NeuralNetwork(input_size, hidden_size, output_size)
+nn.train(X_train, y_train_encoded)
+
+# Save the model
+save_model(nn, '/content/models', 'neural_network_model.pkl')
+```
+
+### Model Loading and Prediction
+
+Load the saved model and predict the orientation of new images.
+
+
+# Load the model
+nn_loaded = load_model('/content/models', 'neural_network_model.pkl')
+
+# Perform predictions
+predictions = nn_loaded.forward(predict_images)
+```
+
+### Visualization
+
+Display the images along with their predicted labels.
+
+
+display_images_with_predictions(name_list, predictions, reverse_label_map, path=data_path, max_images=20)
+```
+
+## Usage
+
+1. **Download the dataset**:
+   - Update the `main_url` to the URL containing the face images.
+   - Run the script to download and save the images to a specified directory.
+
+2. **Preprocess the dataset**:
+   - Use the `resize_image` function to preprocess the images.
+
+3. **Train the model**:
+   - Define the neural network structure.
+   - Train the model using the preprocessed images and labels.
+
+4. **Save the trained model**:
+   - Use the `save_model` function to save the trained model.
+
+5. **Load the model and predict**:
+   - Use the `load_model` function to load the saved model.
+   - Predict the orientation of new face images using the loaded model.
+
+6. **Display predictions**:
+   - Use the `display_images_with_predictions` function to visualize the predictions.
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+## Acknowledgments
+
+- The face images are sourced from [Carnegie Mellon University](https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/).
+

Face_Orientation_Predictor/README.md

@@ -0,0 +1,126 @@
+Sure! Below is a sample description for your README file on GitHub, detailing the task and providing instructions on how to run the code.
+
+---
+
+# Neural Network for Face Orientation Prediction
+
+This repository contains a neural network implementation for predicting the orientation of faces in grayscale images. The model is trained on a dataset of face images, and it can predict four possible orientations: straight, up, left, and right.
+
+## Project Overview
+
+This project involves the following steps:
+
+1. **Data Collection**: Download face images from a specified URL and save them to a local directory.
+2. **Preprocessing**: Resize the images to 32x30 pixels and normalize the pixel values to the range [0,1].
+3. **Model Training**: Train a neural network to classify the orientation of the faces.
+4. **Model Saving**: Save the trained model to a file for future use.
+5. **Model Loading and Prediction**: Load the saved model and use it to predict the orientation of new face images.
+6. **Visualization**: Display the images along with their predicted labels.
+
+## Getting Started
+
+### Prerequisites
+
+- Python 3.x
+- Required Python libraries: `numpy`, `requests`, `beautifulsoup4`, `opencv-python`, `matplotlib`, `Pillow`
+
+Install the required libraries using pip:
+
+```bash
+pip install numpy requests beautifulsoup4 opencv-python matplotlib Pillow
+```
+
+### Data Collection
+
+The script downloads face images from the following URL: [Carnegie Mellon University Face Images](https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/).
+
+```python
+# URL of the webpage
+main_url = "https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/"
+
+# Directory to save downloaded files
+base_directory = "/content/faces"
+
+# Downloading function
+# [Include the code for downloading files here]
+```
+
+### Preprocessing
+
+Resize images to 32x30 pixels and normalize them to the range [0,1].
+
+```python
+def resize_image(image_path):
+    img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
+    resized_img = cv2.resize(img, (32, 30))
+    return resized_img / 255.0
+```
+
+### Model Training
+
+Train the neural network on the preprocessed images and save the model.
+
+```python
+# Define and train the neural network
+input_size = 960  # 32 * 30
+hidden_size = 100
+output_size = 4
+
+nn = NeuralNetwork(input_size, hidden_size, output_size)
+nn.train(X_train, y_train_encoded)
+
+# Save the model
+save_model(nn, '/content/models', 'neural_network_model.pkl')
+```
+
+### Model Loading and Prediction
+
+Load the saved model and predict the orientation of new images.
+
+
+# Load the model
+nn_loaded = load_model('/content/models', 'neural_network_model.pkl')
+
+# Perform predictions
+predictions = nn_loaded.forward(predict_images)
+```
+
+### Visualization
+
+Display the images along with their predicted labels.
+
+
+display_images_with_predictions(name_list, predictions, reverse_label_map, path=data_path, max_images=20)
+```
+
+## Usage
+
+1. **Download the dataset**:
+   - Update the `main_url` to the URL containing the face images.
+   - Run the script to download and save the images to a specified directory.
+
+2. **Preprocess the dataset**:
+   - Use the `resize_image` function to preprocess the images.
+
+3. **Train the model**:
+   - Define the neural network structure.
+   - Train the model using the preprocessed images and labels.
+
+4. **Save the trained model**:
+   - Use the `save_model` function to save the trained model.
+
+5. **Load the model and predict**:
+   - Use the `load_model` function to load the saved model.
+   - Predict the orientation of new face images using the loaded model.
+
+6. **Display predictions**:
+   - Use the `display_images_with_predictions` function to visualize the predictions.
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+## Acknowledgments
+
+- The face images are sourced from [Carnegie Mellon University](https://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-8/faceimages/faces/).
+