Improving Point Cloud Self-Supervised Learning through Object Exchange
The core message of this paper is to introduce a novel self-supervised learning framework for point clouds that leverages an object exchange strategy and a context-aware object feature learning approach to extract robust features that encapsulate both object patterns and contextual information.
3DMambaComplete: A Novel Point Cloud Completion Network Leveraging Structured State Space Model
3DMambaComplete, a novel point cloud completion network, effectively reconstructs complete and high-fidelity point clouds from incomplete and low-quality inputs by incorporating the Structured State Space Model framework.
Efficient and Generic Point Model for Lossless Point Cloud Attribute Compression
The authors propose PoLoPCAC, an efficient and generic lossless point cloud attribute compression method that achieves high compression efficiency and strong generalizability simultaneously by formulating lossless attribute compression as the task of inferring explicit distributions of attributes from group-wise autoregressive priors.
3DMambaIPF: A Scalable and Realistic Point Cloud Denoising Model Leveraging Mamba and Differentiable Rendering
3DMambaIPF introduces a novel iterative point cloud filtering model that leverages the Mamba module for efficient long-sequence modeling and integrates a differentiable rendering loss to enhance the visual realism of denoised geometric structures, enabling superior performance on both small-scale and large-scale point cloud datasets.
Learning Instance-Aware Correspondences for Robust Multi-Instance Point Cloud Registration in Cluttered Scenes
A coarse-to-fine approach to extract instance-aware correspondences for robust multi-instance point cloud registration, which can effectively handle cluttered scenes and heavily-occluded instances.
Comprehensive Survey of Point Cloud Data Augmentation Methods for Deep Learning
This article presents a comprehensive survey of various point cloud data augmentation methods, categorizing them into a taxonomy framework comprising basic and specialized techniques. It evaluates the potentials and limitations of these augmentation methods, serving as a useful reference for selecting appropriate augmentation strategies.
A Highly Efficient and Robust Local Point Cloud Geometry Encoder: VecKM
VecKM is a novel local point cloud geometry encoder that is descriptive, efficient, and robust to noise. It achieves this by vectorizing a kernel mixture representation of the local point cloud, which is proved to be reconstructive and isometric to the original local shape.
Efficient FPGA-Accelerated Point Cloud Registration with PointNet Features
The paper proposes efficient FPGA accelerator cores, PointLKCore and ReAgentCore, for deep learning-based point cloud registration methods that avoid costly feature matching.
Efficient Point Cloud Reconstruction and Denoising via Learned Gaussian Splats and Fine-Tuned Diffusion Features
The authors propose a method to reconstruct point clouds from few images and denoise point clouds from their rendering by exploiting prior knowledge distilled from image-based deep learning models.
Exploring Transfer Learning with Point Transformers: Evaluating Classification Performance on ModelNet10 and 3D MNIST Datasets
Point Transformers, a self-attention based architecture, can effectively capture spatial dependencies in point cloud data and achieve near state-of-the-art performance on various 3D tasks. However, the transfer learning capabilities of these models are limited when the source and target datasets have significantly different underlying data distributions.
© 2024 by Linnk AI