Facial Emotional Recognition With Deep Learning On Pepper Robot

Abstract

There are millions of people in the world, who suffer from life-threatening diseases each year. It has been reported that the magnitude of the disparities of healthcare quality worldwide is still significant. In particular, things can lead to high mortality, such as inadequate doctor-patient ratio and lack of intelligent medical facilities. This situation, however, has been largely alleviated by the rapid developments in fields of Computer Aided Design (CAD) and Robotics. For example, the computer-aided machine can help doctors analyze radiological images and make diagnosis more quickly and accurately; the medical automatic system offers patients a broad and wide access to medical resources in a more effective way. This project is the use of robots for capturing and identifying the patient's expression, through the deep learning of the computer to determine the patient's mental state. The main research works concludes six contents, face detected by photos, resolved by Open Source Computer Vision Library (OpenCV), facial dataset, resolved by Kaggle’s Fer2013, the processing training and testing the facial dataset, resolved by the model Convolutional Neural Network (CNN), showing the result by a bar chart is the next one, resolved by Matplotlib and taking pictures for patients, resolved by one of Pepper’s modules, ALPhotoCapture and face detected by the robot, resolved by one of Pepper’s modules, ALFaceDetection.