Yashaswini
← Back to projects

Facial recognition

This project focuses on recognizing human facial expressions using deep learning techniques. It leverages Convolutional Neural Networks (CNNs) to classify emotions such as happy, sad, angry, and surprised from facial images.

Timeline
Jan 2023 – Apr 2023
Role
Graduate Researcher
Stack
Python, R, pandas, statsmodels, TensorFlow, Keras
Dataset
A combination of publicly available facial emotion datasets (FER-2013, CK+) (daily)

Approach

  • A CNN-based deep learning model was trained on labeled facial expression datasets to extract and learn key facial features.
  • The model was optimized using data preprocessing, augmentation, and performance evaluation metrics to improve classification accuracy.
  • Tuned hyperparameters with rolling-window cross validation to align training with real-world deployment constraints.
  • Use a Convolutional Neural Network (CNN) to extract facial features and classify or match faces based on embeddings.

Highlights

  • Developed and trained a CNN model for multi-class emotion classification.
  • Implemented image preprocessing and augmentation techniques.
  • Achieved high validation accuracy with optimized hyperparameters.