A leap forward in 3D technology by global researchers.
Researchers from The University of Tokyo, RIKEN, South China University of Technology, and Shenzhen Institute of Advanced Technology have collaboratively developed a novel method named GaussianOcc for estimating 3D occupancy. This technique uses Gaussian splatting, an approach that is self-supervised and boasts efficiency in rendering 3D data. Contributors Wanshui Gan and Fang Liu have shared equal credit for this work, while Naoto Yokoya serves as the corresponding author. Their findings have been published on arXiv, a prestigious platform for scientific papers. Additionally, the team has made the practical algorithms available for use, inviting further exploration and development in the field.
This breakthrough contributes significantly to the progression of 3D modeling and representation, facilitating numerous applications across various fields such as virtual reality, gaming, and advanced simulations. The accessibility of both the paper and the code promotes an environment of shared knowledge and collective advancement in computer vision and artificial intelligence, reflecting a growing trend of openness in scientific research.
Read more:
arXiv