Motion capture: Building A 2D virtual character to mimic human performance using a computer vision- based system

Abstract

With the rapid advancement of technology, virtual reality experiences have gained immense popularity, leading to an increased interest in understanding the creation and potential applications of virtual characters. This thesis project introduces an innovative approach to applying motion capture using computer vision-based techniques to generate a two-dimensional virtual character that can stimulate human performance in real-time. By using a computer webcam, the proposed method can track and capture human movements and emotions, which are then translated into a two-dimensional model using Python and C# algorithms. Complementary tools like Clip Studio, Live2D, and Unity are made used to generate the virtual character data and create the virtual environment for them._____ The proposed method is divided into three key phases in order to achieve a fully functional result: (1) the first phase consists of collecting virtual character data, (2) the second phase constructs systems to detect and track body movements, and (3) the last phase integrates these processes into the virtual environment in Unity. Two systems are established in order to give the virtual character the capability to imitate human performance: the Half-body Motion Tracking system and the Facial Expression Detection system. _____ Generally, this thesis project introduces a novel and practical method for motion capture, enabling the development of virtual experience in the digital realm with a simpler, more cost-effective, with efficient approach. It also emphasizes the potential applications of virtual humans in areas such as identity security, the entertainment industry, education, and other areas involving human-computer interaction. With instantaneous capabilities and adaptability, this approach demonstrates the epic influence of virtual humans. Offering new opportunities and enlarging the horizons of human-computer interaction.