Research Interests
Click-based Interactive Segmentation Network for Point Clouds
Interactive segmentation is one meaningful research topic in recent years. Its goal is to obtain segmented objects of interest from interactive information like strokes, bounding boxes, and clicks. These user-selected segmentation results can be used in various downstream tasks, such as medical object detection and annotation of images and videos. In the field of point clouds, the cost of point clouds annotation is higher than that of images, significantly increasing the expenses associated with high-precision 3D map production and other downstream tasks. Moreover, in recent years, there has been a substantial surge in point cloud data due to the widespread use of point cloud mapping equipment and depth cameras. However, traditional labeling methods are incapable of coping with this rapid expansion in point cloud data, presenting challenges in training large network models. So I want to propose more intelligent point cloud interactive segmentation algorithms to streamline the labeling process and reduce costs. My current research focuses on interactive segmentation for point clouds, semantic segmentation for point clouds, instance segmentation for point clouds, and weakly/semi-supervised segmentation for point clouds.
Education
M.Eng., Telecommunications Engineering, Xiamen University, Sept 2019 – June 2022
B.Eng., Telecommunications Engineering, Xiamen University, Sept 2015 – June 2019
Publication
W. Sun et al., "A Click-Based Interactive Segmentation Network for Point Clouds," in IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1-12, 2023, Art no. 5407412, doi: 10.1109/TGRS.2023.3323735.