FacialNet

FacialNet is a deep learning-based Facial Emotion Recognition (FER) system designed to classify human emotions from visual data. The project features dedicated pipelines for both static image analysis and dynamic video prediction, leveraging powerful Convolutional Neural Network (CNN) architectures to achieve high accuracy.

Key Features

• Multi-Model Architecture: Implements both ResNet and DenseNet architectures to extract nuanced facial features and compare performance metrics.
• Static Image Prediction: A dedicated module for processing and classifying emotions from standalone images.
• Dynamic Video Prediction: A robust pipeline capable of analyzing video streams frame-by-frame for continuous emotion tracking.
• Research-Driven Notebooks: Includes comprehensive Jupyter Notebooks detailing the model training, evaluation, and inference processes.

Tech Stack

• Programming Language: Python
• Machine Learning Frameworks: PyTorch / TensorFlow
• Computer Vision: OpenCV (for video/image processing)
• Environment: Jupyter Notebook

Repository Structure

FacialNet/
├── Image prediction/         # Scripts and assets for static image emotion recognition
├── Video Predicition/        # Scripts and assets for continuous video emotion tracking
├── densenet_model.ipynb      # Training and evaluation pipeline using DenseNet
├── resnet_model.ipynb        # Training and evaluation pipeline using ResNet
├── requirements.txt          # Project dependencies
└── .gitignore                # Git ignore rules