Case Study
Category:
Computer Vision & Human-Computer Interaction
Impact:
10 Weeks | Fast Multi-Class Detection
Understanding human emotions through facial expressions is vital for applications in research, user experience testing, education, and human-computer interaction. Manual observation is subjective and inefficient, while existing tools often lack flexibility and real-time usability. To address this gap, I created a web-based facial emotion recognition system capable of analyzing facial expressions across multiple categories such as happy, sad, angry, and others. The application provides a lightweight, accessible solution with a user-friendly interface for uploading images and obtaining fast predictions.
Implemented preprocessing techniques for facial images, including resizing and normalization, to ensure consistent inputs.
Trained a Convolutional Neural Network (CNN) using Keras on seven labeled emotion classes.
Built a lightweight Flask API delivering JSON outputs for real-time predictions with minimal latency.
Developed a browser-based UI (HTML/JavaScript) allowing users to upload images and visualize predicted emotions.
