Point Cloud
Point clouds are collections of 3D data points representing objects or scenes, primarily used for tasks like 3D reconstruction, object recognition, and autonomous navigation. Current research focuses on improving the efficiency and robustness of point cloud processing, employing techniques like deep learning (e.g., transformers, convolutional neural networks), optimal transport, and Gaussian splatting for tasks such as registration, completion, and compression. These advancements are crucial for applications ranging from robotics and autonomous driving to medical imaging and cultural heritage preservation, enabling more accurate and efficient analysis of complex 3D data.
Papers
RecNet: An Invertible Point Cloud Encoding through Range Image Embeddings for Multi-Robot Map Sharing and Reconstruction
Nikolaos Stathoulopoulos, Mario A. V. Saucedo, Anton Koval, George Nikolakopoulos
Mitigating Prior Shape Bias in Point Clouds via Differentiable Center Learning
Zhe Li, Jinglin Zhao, Zheng Wang, Bocheng Ren, Debin Liu, Ziyang Zhang, Laurence T. Yang
MixSup: Mixed-grained Supervision for Label-efficient LiDAR-based 3D Object Detection
Yuxue Yang, Lue Fan, Zhaoxiang Zhang
Dynamic Prototype Adaptation with Distillation for Few-shot Point Cloud Segmentation
Jie Liu, Wenzhe Yin, Haochen Wang, Yunlu CHen, Jan-Jakob Sonke, Efstratios Gavves
LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detection
Sifan Zhou, Liang Li, Xinyu Zhang, Bo Zhang, Shipeng Bai, Miao Sun, Ziyu Zhao, Xiaobo Lu, Xiangxiang Chu