Detection of emotional expressions and application to real-time images with faster R-CNN deep learning method

In this study, using the Faster R-CNN deep learning method, emotional expressions of anger, happy, neutral, fear, sadness, disgust and surprise are determined from portrait face (frontal) images. A social life-oriented decision/support application is developed over real-time images with the detected expressions. In the study, first of all, seven emotional expressions are labeled from the portrait face images that formed the data set. Then, the tagged images are digitized and the learning/training process of seven emotional expressions is completed on the model developed with the Faster R-CNN deep learning algorithm. Finally, a decision/support application is developed that tests the success of seven emotional expressions on real-time images with the Python program. As a result of the validation with 350 test data, the educational success results were anger %92, happiness %98, neutral %98, fear %94, sadness %94, disgust %86 and surprised %94.