ACADEMIC FACIAL EXPRESSION DETECTION USING CNN-RNN(CLASSIFICATION) and CNN-LSTM(RECOGNITION) DEEP LEARNING

ACADEMIC FACIAL EXPRESSION DETECTION USING CNN-RNN(CLASSIFICATION) and CNN-LSTM(RECOGNITION) DEEP LEARNING

Shruti Sahare, Prof. Snehal Rathi

BRACT’S Vishwakarma Institute of Information Technology, Survey No. 3/4, Kondhwa (Budruk), Pune: 411048

ABSTRACT –

Online education systems have evolved during the past two years as physical teaching has been blocked due to the global pandemic, so this project plays an important role in understanding student’s academic emotions such as Boredom, Confusion, Engagement, and Frustration during E-learning. We can track the progress of students using this technology for better understanding and communication between teachers and students. Our technology is basically focused on the deep learning approach, for classification we are using Convolutional Neural Network (CNN) along with Recurrent Neural Network (RNN,) and for recognition we are using Convolutional Neural Network (CNN) along with Long-Short Term Memory (LSTM).

In this project we have successfully created the real-time model with the accuracy ranging between 50-60 because the dataset isn’t made for recognition purposes and to achieve a higher accuracy, we planned to create our own dataset with a group of students and plan to go further with this research.

Index Terms - Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), Graphics Processing Unit (GPU), Recurrent Neural Network (RNN), Software Development Life Cycle (SDLC), Unified Modelling Language (UML), High Definition (HD), Red Green Blue (RGB)