Facial-Emotion-Recognition-System
The Facial Emotion Recognition System is a PyTorch-based computer vision project that classifies facial expressions into seven categories: happy, neutral, surprise, sad, angry, fear, and disgust. Trained on the FER-2013 dataset, it supports two model backbones: a lightweight custom CNN and a ResNet18-based transfer learning architecture. The repository is organized with a modular production-style structure, including separate directories for source code, training scripts, evaluation outputs, saved model artifacts, configuration files, and unit tests. Key features include reproducible training workflows via CLI commands, evaluation utilities that generate classification reports and confusion matrices, and inference tooling for predicting emotions from input images. The project demonstrates ML engineering practices such as artifact management, dependency consistency, and separation between experimentation and application logic. It serves as a portfolio piece showcasing end-to-end image classification beyond a s