自动驾驶感知
文章平均质量分 95
陈同学_alex
这个作者很懒,什么都没留下…
展开
-
深度估计的地面假设
则图像中每个像素对应的3D点,等于从相机光心发射的、通过该像素的射线与地面的交点。求解该交点,并将其投影到相机坐标系,即可得到每个像素距离地面的深度。平面外某条射线的起点。假设已知相机到地面的位姿。求:射线 和平面的交点。已知:空间平面上的点。原创 2023-09-13 23:08:43 · 195 阅读 · 0 评论 -
MVSNet,CVP-MVSNet论文阅读和代码解析
提出一种基于cost volume的神经网络,用于多视图图像的深度推断。我们证明了以一种coarse-to-fine的方式构建cost volume金字塔,而不是以固定分辨率构建cost volume,可以得到一个紧凑、轻量级的网络,并允许推断高分辨率深度图,以获得更好的重建结果。为此,首先以图像的最粗分辨率在整个深度范围内对前平行平面(fronto-parallel planes)进行均匀采样,以此为基础构建cost volume。然后,给定当前深度估计,我们在像素深度残差上迭代构建新的cost volu原创 2023-09-11 12:43:14 · 290 阅读 · 1 评论 -
3D Dynamic Scene Graphs论文阅读
3D Dynamic Scene Graphs: Actionable Spatial Perception with Places, Objects, and HumansAntoni Rosinol, Arjun Gupta, Marcus Abate, Jingnan Shi, Luca CarloneLaboratory for Information & Decision Systems (LIDS)Massachusetts Institute of TechnologyRSS2原创 2022-05-22 16:28:18 · 757 阅读 · 1 评论 -
CubeSLAM论文阅读
CubeSLAM: Monocular 3-D Object SLAMShichao Yang and Sebastian Scherer摘要In this paper, we present a method for single image three-dimensional (3-D) cuboid object detection and multiview object simultaneous localization and mapping in both static and dyn原创 2022-05-11 21:20:11 · 1061 阅读 · 0 评论 -
光流估计算法RAFT的论文和代码阅读
RAFT论文和代码阅读RAFT: Recurrent All-Pairs Field Transforms for Optical FlowZachary Teed and Jia DengECCV2020 best paper摘要We introduce Recurrent All-Pairs Field Transforms (RAFT), a new deep network architecture for optical flow. RAFT extracts per-pixel fe原创 2022-04-08 22:12:00 · 3572 阅读 · 1 评论 -
PointNet++论文和代码阅读
PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric SpaceCharles R. Qi Li Yi Hao Su Leonidas J. GuibasStanford UniversityAbstractFew prior works study deep learning on point sets. PointNet is a pioneer in this direction. However, b原创 2022-04-03 19:28:27 · 4149 阅读 · 0 评论