点云论文阅读
疯子_Zhang
暂无简述
展开
-
Transformer在点云上的应用
快速浏览了这两篇文章,方法都比较简单,一些思想都在前面的工作中或多或少都出现过。值不值的读,能不能中不好说。第二篇文章的方法还是很涨点的。1 PCT: Point Cloud Transformer2 Point Transformer...原创 2020-12-24 20:38:04 · 2024 阅读 · 1 评论 -
论文泛读:Distribution aligning refinery of pseudo-label for imbalanced semi-supervised learning
论文泛读:Distribution aligning refinery of pseudo-label for imbalanced semi-supervised learning原创 2020-12-16 21:48:26 · 1071 阅读 · 0 评论 -
3D 论文阅读SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images
浏览SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images在解决什么问题?面临的困难本文的方法Signed distance functionsRendering Implicit 3D Surfaces(I donot care)在解决什么问题?如何从单幅图像(仅仅利用轮廓)学习3D形状。已有的全监督的方法需要成对的注释(每个图像都有对应的3D模型),显然这很费劲;还有一些方法是用多视角的轮廓,多视角能原创 2020-11-28 12:57:48 · 956 阅读 · 0 评论 -
点云语义分割:论文阅读简记 -Spatial Transformer Point Convolution
点云语义分割:Spatial Transformer Point Convolution论文阅读你好! 这是你第一次使用 Markdown编辑器 所展示的欢迎页。如果你想学习如何使用Markdown编辑器, 可以仔细阅读这篇文章,了解一下Markdown的基本语法知识。新的改变我们对Markdown编辑器进行了一些功能拓展与语法支持,除了标准的Markdown编辑器功能,我们增加了如下几点新功能,帮助你用它写博客:全新的界面设计 ,将会带来全新的写作体验;在创作中心设置你喜爱的代码高亮样式,M原创 2020-10-26 10:37:41 · 590 阅读 · 0 评论 -
3D 论文阅读 PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding简记
PointContrast: Unsupervised Pre-training for3D Point Cloud Understanding摘要PointContrast Pre-training3D Point Cloud Understanding)摘要简单记一下Charles R. Qi的新作 PointContrast: Unsupervised Pre-training for3D Point Cloud Understanding。点云的无监督的对比学习作为预训练以提升下游任务的性原创 2020-07-28 07:52:39 · 1509 阅读 · 0 评论 -
3D点云论阅读:ShellNet:Efficient Point Cloud Convolutional Neural Networks using Concentric Shells Statics
论文:http://openaccess.thecvf.com/content_ICCV_2019/papers/Zhang_ShellNet_Efficient_Point_Cloud_Convolutional_Neural_Networks_Using_Concentric_Shells_ICCV_2019_paper.pdf源码:https://github.com/hkust-vgd...原创 2020-03-02 09:54:53 · 1152 阅读 · 0 评论 -
3D点云实例分割
3D点云实例分割3D语义分割区分场景中各类对象,3D实例分割区分场景中各类别中的各种个体。近两年来,3D实例分割的关注度越来越高,相应的方法也被接连提出。众多方法的思想主要分为两类:基于候选区域的实例分割(proposal-based)和免候选区域的实例分割(proposal-free)。其中,proposal-based先获取场景中的感兴趣的候选区域,如:3D bounding boxes等...转载 2020-01-07 14:52:15 · 10897 阅读 · 14 评论 -
点云的实例分割和语义分割
SGPN [CVPR 2018]:点云的实例分割与物体检测。(SGPN: Similarity Group Proposal Network for 3D Point Cloud Instance Segmentation。RSNet [CVPR 2018]:点云的语义分割。(Recurrent Slice Networks for 3D Segmentation on Point Cloud...转载 2019-12-27 19:36:33 · 12947 阅读 · 1 评论 -
Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks
Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks(地基打得好,突破上限不会倒)闲来无事,找篇论文粗读一下:Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks。初看题目Break the Ceil...原创 2019-12-27 09:22:58 · 1222 阅读 · 1 评论 -
3D论文阅读:JSIS3D:Joint Semantic-Instance Segmentation...
论文地址:https://arxiv.org/abs/1904.00699代码(论文中有地址)环境: CUDA 10.0; Pytorch 1.0;Python 3.5+;Pytorch 0.4.0+之前知道有一篇联合训练的文章,一直没有看,最近粗略阅读并记录一下。。。...原创 2019-11-27 10:38:21 · 1672 阅读 · 9 评论 -
三维点云论文阅读:Graph Attention Convolution for Point Cloud Segmentation
本文是方法的概览,下文不包含可舍弃条件随机场的理论分析部分,请查阅原文。原创 2019-10-22 16:38:35 · 1136 阅读 · 1 评论 -
三维目标检测论文阅读:RoarNet: A Robust 3D Object Detection based on RegiOn Approximation Refinement
RoarNet: A Robust 3D Object Detection based on RegiOn Approximation Refinement (速读)论文地址:https://arxiv.org/pdf/1811.03818.pdf代码地址:https://github.com/Kiwoo/RoarNet官方介绍主页:https://sites.google.com/b...原创 2019-04-01 21:10:52 · 1413 阅读 · 0 评论 -
三维目标检测论文阅读:Frustum PointNets for 3D Object Detection from RGB-D Data
Frustum PointNets for 3D Object Detection from RGB-D Data(有代码,级联的方法)目录Frustum PointNets for 3D Object Detection from RGB-D Data简要说明Contributions(不对应原文)具体方案实验代码学习简要说明问题:通常的三维目标(用...原创 2019-04-08 20:45:11 · 4427 阅读 · 1 评论 -
三维目标检测论文阅读:PIXOR: Real-time 3D Object Detection from Point Clouds
PIXOR: Real-time 3D Object Detection from Point Clouds论文链接:http://www.cs.toronto.edu/~wenjie/papers/cvpr18/pixor.pdf代码链接:https://github.com/Yachao-Zhang/PIXOR摘要:本文是18年CVPR,时间比较久了网上解读版本很多,本文只是记录一...原创 2019-05-07 13:52:46 · 3013 阅读 · 0 评论 -
三维目标检测论文:Deep Hough Voting for 3D Object Detection in Point Clouds
3D目标检测框架VoteNetCharles R. Qi,Or Litany,何恺明,Leonidas J. Guibas等当前主流的3D目标检测方法,很大层度上受2D检测器的影响。充分利用2D检测器提取3D建议框,或将3D点云转换为规则的网格或利用2D图像提取锥形区域可参阅(https://blog.csdn.net/u011507206/article/details/89106892...转载 2019-04-26 09:54:36 · 10253 阅读 · 6 评论 -
三维目标检测论文阅读:VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection
VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection理解了PointPillars这篇文章后,看这篇文章就清晰多了,这个清晰表现在两个方面,1)PointPillars这篇文章具体是如何将点云变为伪图像过程变得清晰了,2)VoxelNet这篇文章多用公式表达并且很详细,读起来是比较容易懂的。建议先看本文再...原创 2019-06-05 14:03:08 · 2298 阅读 · 0 评论 -
【ICLR2019】Poster 论文汇总
【ICLR2019】Poster 论文汇总转自:https://blog.csdn.net/u014636245/article/details/88825615ICLR2019Poster 文章汇总, 共478 papersConvolutional Neural Networks on Non-uniform Geometrical Signals Using Eucli...转载 2019-06-05 20:53:35 · 2185 阅读 · 0 评论 -
三维目标检测论文阅读:PointPillars: Fast Encoders for Object Detection from Point Clouds
PointPillars: Fast Encoders for Object Detection from Point Cloudscode ID: https://github.com/nutonomy/second.pytorch目录PointPillars: Fast Encoders for Object Detection from Point Clouds概...原创 2019-06-04 17:53:36 · 14425 阅读 · 2 评论 -
ICCV2019点云论文 ResGCN-Can GCNs Go as Deep as CNNs
ICCV 2019 Oral论文速读残差图卷积网络-Can GCNs Go as Deep as CNNsArxivpaper:https://arxiv.org/abs/1904.03751Github:https://github.com/lightaime/deep_gcns...原创 2019-08-03 10:48:19 · 3306 阅读 · 6 评论 -
三维目标检测论文阅读:Deep Continuous Fusion for Multi-Sensor 3D Object Detection
题目:Deep Continuous Fusion for Multi-Sensor 3D Object Detection来自:Uber: Ming LiangNote: 没有代码,主要看思想吧,毕竟是第一篇使用RGB feature maps 融合到BEV特征中;KITTI 3D目标检测数据可视化参阅:https://blog.csdn.net/u011507206/article...原创 2019-03-30 10:52:40 · 2963 阅读 · 0 评论