Xuran Zhao

Title: Innovations of Xuran Zhao in Depth Map Generation

Introduction

Xuran Zhao is a notable inventor based in Zhejiang, China. He has made significant contributions to the field of computer vision, particularly in the area of depth map generation. His innovative approach utilizes convolutional neural networks to enhance the quality of 2D-3D conversions in television works.

Latest Patents

Xuran Zhao holds a patent for a method titled "Method for generating spatial-temporally consistent depth map sequences based on convolution neural networks." This patent outlines a systematic approach that includes several key steps. First, a training set is collected, which consists of continuous RGB image sequences and their corresponding depth map sequences. Next, each image sequence undergoes spatial-temporal consistency superpixel segmentation, leading to the establishment of spatial and temporal similarity matrices. The invention also details the creation of a convolutional neural network that incorporates a superpixel depth regression network and a spatial-temporal consistency condition random field loss layer. The final steps involve training the network and recovering depth maps from RGB image sequences of unknown depth, thereby overcoming limitations associated with traditional depth recovery methods.

Career Highlights

Xuran Zhao is affiliated with Zhejiang Gongshang University, where he contributes to research and development in the field of computer science. His work focuses on advancing techniques that improve the accuracy and efficiency of depth map generation, which is crucial for various applications in multimedia and virtual reality.

Collaborations

Xuran Zhao collaborates with Xun Wang, who is also involved in similar research areas. Their partnership enhances the potential for innovative breakthroughs in depth map technology.

Conclusion

Xuran Zhao's contributions to depth map generation through convolutional neural networks represent a significant advancement in the field of computer vision. His innovative methods pave the way for improved 2D-3D conversions, showcasing the importance of research in this area.