Abstract:
Emotions of a human plays an important role in human life shaping the thoughts, behaviours
and relationships of a person. It is helpful in responding in different situations in life
appropriately and quickly like in taking decisions and solving problems. Because of negative
emotions people feel some of them take negative decisions and actions which harms towards
their lives. With the modern busy lifestyle many people struggle with depression and stress due
to the work load they have to handle in both in their work life and personal life. So having the
capability to detecting emotions and understanding patterns through it can make a person to
prevent being negative and live a positive mindset in life.
This research Focuses on developing a strong model which deals with accurately detecting and
classifying human emotions from facial expressions. The methodology basically functions with
CNN architecture and transfer learning approach using MobileNet model. This approach
mostly benefited from the existing knowledge provided with MobileNet contributing to the
model's efficiency and accuracy the system was praying to identify nine distinct emotions
angry, confused, disgusted, fearful, happy ,neutral, sad, shy and surprised.
This model achieved 82% accuracy in recognizing these emotions demonstrating capability at
integrating such technologies into mental healthcare strategies. The findings highlight the
potential of automated facial motion analysis as an important tool for early detection and
therapeutic interventions for various mental health conditions
