A Portfolio of
Data-Driven Solutions

Facial Expression Classification on the LFW Dataset

This project presents a novel approach to classifying facial expressions, specifically focusing on smiling and not smiling faces, using the Labeled Faces in the Wild (LFW) dataset. The LFW dataset comprises 13,143 RGB face images, traditionally categorized based on individual identities. Leveraging the deep funneled version of this dataset, we integrated an external attributes dataset that scores various facial attributes to enhance our classification task. The model achieved a validation accuracy of 99.25%, demonstrating its effectiveness in this novel classification task.