I worked on a digital billboard platform where the main challenge was trust in listings. My focus was on shaping a simple verification flow that made the process more secure and reliable for users

Project Abstract:  Facial Emotion Recognition (FER) System

This research presents a critical review and comparative analysis of Facial Expression Recognition (FER) techniques, focusing on advancements in machine learning (ML) and deep learning (DL) from 2020 to 2024. Recognizing the increasing necessity for sophisticated emotional interpretation in behavioral analysis, our study emphasizes developing versatile and efficient models applicable across diverse digital environments, particularly for analyzing expressions in children.

We advocate for a balanced, high-efficiency approach for static datasets, integrating transfer learning methods (utilizing VGG16 for Convolutional Neural Networks) with robust ensemble techniques (combining Gradient Boosting and Support Vector Machine). This strategy was designed to optimize the trade-off between model complexity and data requirements while maximizing accuracy.

The integrated methodology proved highly effective, demonstrating that the Ensemble Model achieved a leading performance accuracy of 85.0%. This work contributes a validated, optimized framework designed to enhance the reliability and real-world practicality of emotional detection systems for specialized applications.

For my final year project, I led the development of a system that recognizes children’s emotions in classroom settings to support their emotional well-being. I worked on the core functionality using deep learning models and later designed child-friendly screens with a focus on simplicity, usability, and positive interaction. Alongside the project, I co-authored a detailed research paper documenting our approach, experiments, and results.