A Review of 3D Avatar Reconstruction for Virtual Conferencing

3D avatar reconstruction technology has been extensively studied in recent years and has been used to enhance the sense of interaction in virtual conferencing. 3D avatar reconstruction technology enables more realistic communication and interaction by transforming the attendees’ real appearance into a digital 3D avatar and entering the meeting on behalf of the attendees. Although significant progress has been made in 3D avatar reconstruction technology, it remains a challenge to recreate a 3D avatar with more detail to enter the conference in place of a human. Through an in-depth analysis of the current situation of 3D avatar in virtual conference, traditional 3D avatar reconstruction pipeline, and two 3D avatar reconstruction methods based on deep learning, this paper finds that the deep learning-based reconstruction method has the obvious advantages of simple equipment requirements, fast reconstruction speed, and high reconstruction accuracy. This is highly compatible with the demand for enhancing the realism of 3D avatars and performing facial and body pose reconstruction and other details to enhance the virtual conferencing experience. Therefore, this paper provides a useful direction for enhancing the quality of 3D avatars in virtual conferencing. By adopting a deep learning-based approach, accurate reconstruction of participants’ appearance and movements can be more effectively realized, thus further enhancing the realism and interactivity of virtual conferencing. This is of great theoretical and practical significance in meeting the expectations of virtual conferencing participants for high-quality 3D avatars and prompting the development of virtual conferencing technology in a more advanced and detailed direction.