This project involves building and training a Convolutional Neural Network (CNN) using Keras from scratch. The aim is to recognize facial expressions using a dataset consisting of 48x48 pixel grayscale images of faces categorized into seven emotion classes: Angry, Disgust, Fear, Happy, Sad, Surprise, and Neutral.
- Building and training a CNN in Keras from the ground up.
- Processing and utilizing 48x48 pixel grayscale images of faces.
- Employing OpenCV to automatically detect faces in images and draw bounding boxes around them.
- Training, saving, and exporting the CNN model.
- Implementing real-time facial expression recognition on video and image data by serving the trained model to a web interface.
- Implementing Convolutional Neural Networks (CNNs) using Keras.
- Image processing and manipulation for facial recognition.
- Real-time application integration by serving the trained model to a web interface.
The dataset used in this project is sourced from the ICML 2013 Workshop - Competition in Facial Expression Recognition challenge on Kaggle. It comprises a collection of facial expression images labeled into seven categories.