Fast Reconstruction and Real-time Rendering of Immersive Light Field Video

该项目旨在开发新的基于学习的方法，用于从多视图2D视频重建和渲染3D沉浸式视频。该项目预计将在数据挖掘，多媒体，模式识别和深度学习领域产生新的知识。该项目的预期成果包括表示3D视频的新深度神经网络，高保真视频渲染的神经方法以及高效的3D视频重建和渲染算法。这将为各种实际应用提供显着的好处，例如自动驾驶，虚拟现实，医疗保健，先进制造和许多其他3D应用。

This project aims to develop new learning-based methods for reconstructing and rendering 3D immersive videos from multi-view 2D videos. The project expects to generate new knowledge in the areas of data mining, multimedia, pattern recognition and deep learning. Expected outcomes of this project include new deep neural networks to represent 3D videos, neural methods for high-fidelity video rendering and efficient 3D video reconstruction and rendering algorithms. This should provide significant benefits to a diverse range of practical applications, such as autonomous driving, virtual reality, healthcare, advanced manufacturing, and many other 3D applications.