Application: Emotion / Facial Expression Recognition with OpenCV

Опубликовано: 17 Январь 2025
на канале: ICT UoM

322

By Vinasha Chellapermal CVBPR S1 Application23

This application is used to detect emotions or facial expressions on a person’s face. Emotions and facial expressions determine how a person is feeling at a point in time.
In order to detect emotions or expressions on faces, a deep learning approach is normally used by training a Convolutional Neural Network (CNN) on different thousands of facial expression images and after the training it would be able to detect facial expressions when presented with a
new image.
This application uses a deep learning classifier that is loaded to the OpenCV DNN module. The model was trained using MS Cognitive Toolkit and then converted to ONNX (Open neural network exchange ) format.
The model was trained on FER+ dataset, FER dataset is normally used as the standard dataset for emotion recognition task but in FER+, each image has been labeled by 10 crowd-sourced taggers, which thus provides a higher quality of ground truth label for still image emotion than the original FER labels.
The model can detect 8 different kinds of emotions namely: neutral, happiness, surprise, sadness, anger, disgust, fear, and contempt.