Measuring Happiness among Employees in the Manufacturing Industry using Facial Expression Recognition
Happiness in the workplace has been widely discussed throughout the years as it plays a critical role in the success of an organization. However, emotions are often a neglected topic in the workplace despite emotions being the true reflection of inner self, which is a source of truth for employee happiness. Surveys that have been conducted on workplace happiness are mostly traditional survey questions which can be altered depending on what actions have been taken after the results are received by the management. Nowadays with the help of machine learning, facial expressions are able to be captured as a measurement of happiness. To find out the baseline of facial expressions measurement in the industries, the manufacturing industry which ranked as one of the least happy was chosen. In this paper, facial expression recognition is proposed as a survey tool to determine the true expressions from employees and also to determine happiness levels in manufacturing industry at different timeframes. Haar Cascade is applied to determine the facial position in the video frames captured from the webcam, and CNN has been applied to utilize the facial expression recognition model training and prediction. JAFFE and TFEID facial expression datasets are applied to train and test the CNN model. The facial expressions of happy will then be captured into the database for analysis purpose, and to provide helpful insights for the organization to better understand employee happiness throughout the day and improve the environment or behaviour if required. This study was conducted during the COVID-19 pandemic, and consists of the findings of happiness levels compared between 3 periods of online recorded meetings from the employees. The study concludes that the morning session is the happiest at 60% of the time, and least happy during the afternoon session at 38% of the time.
Keywords-Facial Expression Recognition; CNN; Happiness in workplace;Online meetings