Emotion detection through Images with ESP32-CAM and artificial intelligence

Abstract

This project was born with the idea of being able to discern in real time and almost instantly what emotion is reflected on the face of a person — referred to as the “user” — when captured by a camera. To achieve this, we decided to use an ESP32-CAM device, a low-cost microcontroller with a built-in camera and Wi-Fi, and create a functional prototype capable of detecting emotions, as the main aim. Starting with the categories happy and sad, the project aims to evaluate its potential for interactive and assistive applications. This strategy requires us to create a small collection of facial images, using both our own photographs and existing files found online(FER2013), preprocessing them with OpenCV by converting to greyscale, resizing to 48×48 pixels and applying data augmentation. This strategy requires us to create a small collection of facial images, using both our own photographs and existing files found online(FER2013), preprocessing them with OpenCV by converting to greyscale, resizing to 48×48 pixels and applying data augmentation. In addition to emotion recognition, the system also integrates a facial identification module based on Haar Cascade classifiers for face detection and Local Binary Patterns Histograms (LBPH) for user recognition. This allows the system to identify the registered user — in this case, “Judit” — and display their name alongside the detected emotion in real time.