Personality prediction system based on physiognomy using face recognition

the human face is a wealthy source of information, with distinct features such as the eyes, nose, and mouth offering a wealth of data. Facial feature recognition algorithms leverage this wealth of information to distinguish one person from another, making it a powerful tool with many applications. This technology has found its way into numerous aspects of our daily lives, from unlocking smartphones with a simple glance to enhancing security in public spaces. Even more, a person's personality traits can be predicted from their outer appearance according to physiognomy as each shape of facial features can tell a lot about a person's personality; this work addresses this by proposing a novel dataset created for facial feature recognition based on their shape and color, specially designed for the YOLO object detection models; the final dataset contains 2,116 images with Over 10K annotations with six classes, namely blue eye, brown eye, rounded nose, rounded eyebrows, pointy nose, and straight eyebrows, employed in all version of the YOLOv8 object detection model, experiment result shows that the small version of YOLOv8 performed the best based on mAP of 89.9%, this work also proposed a modified lightweight version of YOLOv8 with only 211 layers which outperformed the best model by 1.4% as it reaches 91.3%mAP on the proposed dataset.