QUALITATIVE RECOGNITION STUDY OF EMOTIONS IN REAL TIME FOR CUSTOMER SERVICE USING DEEPLENS FACE DETECTION

Abstract

At present, a growing interest in the emotional state that employees present in customer service within organizations and
companies is a fundamental point that helps guiding the behavior and thought processes in order to achieve good contact with
those who request their services. This study, which has mainly a qualitative characteristic, was carried out with the staff working
in the administrative area of the Educational Unit “Santa Maria Goretti”, Ecuador. Within the framework of their job
performance, this study aims to recognize the state of excitement of the employee. The captured image was processed giving rise
to the recognition of emotions through facial features using a 4-layer convolutional network. The result of the sampling is
subsequently presented, obtaining emotions in the range from which it presents greater percentage weighting to the lowest.
Through the training process it was allowed to capture physical features in real time using the DeepLens face detection
application, we worked with faces at different angles, this application emphasizes three main modules, these are namely, 1) face
detection, 2) removal of features, and 3) classification of expressions. Before performing the task of emotion recognition, the
facial recognition software detected the face and a series of key points as eyes, lips, eyebrows. and cheeks, resulting in the
capture of employee’s excitement, and the percentages of seven emotions, disgust, surprise, fear, angriness, happiness, sadness,
and neutral.