Personality Prediction based on facial features

The ability to predict someone's personality can have numerous applications, ranging from personalized recommendations to identifying mental health issues.In this project, we aim to explore the relationship between facial features and personality traits and develop a machine learning algorithm to predict personality based on facial features. Research has focused on various approaches, including self-assessments, behavioral observations, and physiological measurements. However, these approaches have their limitations, and their accuracy is often questioned.

Dataset

We collected primary data from college students using Google Forms. A total of 138 volunteers participated, aged between 18 to 25 years, with 7 females and the rest males. The form had 38 questions related to five personality traits and an image section where volunteers uploaded their frontal facial image without glasses. Each question was rated on a scale of 0 to 5. The data is stored on a Google Drive accessible only to team members to ensure privacy. However, since there was no invigilator present during data collection, there may be potential biases in the data.

dataset

Challenges Faced Low accuracy rate of 43 percent due to limited dataset size. Lack of expertise in facial feature extraction affecting the precision of feature extraction. Absence of a personality expert impacting the accuracy of the questionnaire design. Questionnaire not filled in the presence of an invigilator, potentially leading to errors in responses. Missing facial pictures from some volunteers resulting in reduced sample size and potential bias. Accuracy reached to 66.67% via alternative approach and techniques.