OTT 이미지에서 인공지능을 활용한 배우의 감정 표정변환 기술

초록

The purpose of this research is to develop an actor's emotional facial expression conversion technology using artificial intelligence to enhance the viewing experience of video contents used in OTT (Over-The-Top) services. In this work, we propose an emotional image conversion rendering algorithm architecture based on convolutional neural network (CNN) by identifying the actor's three-dimensional facial features that combine computer vision and deep learning techniques to the image extracted as a three-dimensional vector value image in the existing deepfake-based technology. By doing so, we aim to detect and analyze the actor's facial expressions in real time and convert them into optimal emotions for the narrative. In addition, this research aims to integrate into OTT platforms to collect and analyze users' reactions and feedback in the future, and above all, to convey the actor's poor performance to the viewer as a means of conveying meaning in a deeper way through AI-based video emotional expression conversion technology. Although this research has some problems, it is ultimately expected to provide viewers with a richer and more lively viewing experience, and will contribute to various entertainment fields in OTT.

키워드