A Proposal for Deep Learning Based Emotion Identification

Emotion detection plays a vital role in understanding human behavior and enhancing human computer interaction. This study presents the development and evaluation of an Emotion Detection Tool utilizing machine learning techniques. The tool aims to classify emotions from facial expressions in real-time video streams. The methodology involved using convolutional neural networks (CNNs) for feature extraction and emotion classification. A publicly available dataset consisting of facial images indexed with six basic feelings is used for testing the proposed model. The implementation includes real-time video capture, facial landmark detection, and emotion inference. The CNN architecture is explored and compared for their performance in feeling identification. This study uses various evaluation measure parameters to judge the model's strength. The results demonstrate promising accuracy in recognizing basic emotions from facial expressions, with the highest accuracy achieved for happiness and surprise.