Pipeline for Development of 3 Dimensional motion animation using the 2 Dimensional video

摘要

The main idea of this project is to propose an efficient set of techniques to perform the process of motion capture. Motion capture is basically a process of capturing the motion of a person and using it to transfer it to a 3D model created to the user's desire. The process basically obtains this motion by making the actor wear a motion capture suit which contains markings on their body joints and the changes in the position of these markers are used to obtain the motion. This project has established a set of techniques to make this process easier and cost-efficient. Fabricating a mobile robot that can follow the actor and record the video by keeping him in the required position in the video frame. Then the recorded video is inputted to several deep learning models which in turn can give us the 3D pose of the actor. This motion can be transferred to the required 3D model. Thus, avoiding the use of the mocap suit and multiple cameras to get the motion from the actor.