单阶段6D对象姿势估计
单阶段6D对象姿势估计
以下文章来源于计算机视觉工坊,作者康斯坦奇
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003649606-1236516577.png)
论文题目:Single-Stage 6D Object Pose Estimation
一、简介
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003705082-821635853.png)
二、方法
2.1 密集点位姿估计
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003705870-1513109048.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003707152-2110888578.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003707639-1949652444.png)
2.1.1 密集点的属性
2.1.2 网络架构
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003710814-751284966.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003711904-918005299.png)
2.2单阶段6D对象姿态估计
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003712510-1782819435.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003713237-1789487348.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003713616-193348954.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003714050-142450400.png)
三、实验
3.1 合成数据
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003715075-2016558657.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003718145-2110684423.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003720595-1375855576.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003722999-1207033859.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003724659-1659794644.png)
3.2 真实数据
3.2.1 遮挡数据及比较结果
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003726534-1993648700.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003727830-1995622383.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003728548-746882997.png)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003745088-58158448.png)
3.2.2 YCB-Video 结果
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003746018-2052111986.png)
3.3 限制
四、结论
重磅!3DCVer-学术论文写作投稿 交流群已成立
扫码添加小助手微信,可申请加入3D视觉工坊-学术论文写作与投稿 微信交流群,旨在交流顶会、顶刊、SCI、EI等写作与投稿事宜。
同时也可申请加入我们的细分方向交流群,目前主要有3D视觉、CV&深度学习、SLAM、三维重建、点云后处理、自动驾驶、CV入门、三维测量、VR/AR、3D人脸识别、医疗影像、缺陷检测、行人重识别、目标跟踪、视觉产品落地、视觉竞赛、车牌识别、硬件选型、学术交流、求职交流等微信群。
一定要备注:研究方向+学校/公司+昵称,例如:”3D视觉 + 上海交大 + 静静“。请按照格式备注,可快速被通过且邀请进群。原创投稿也请联系。
![3D视觉工坊小助理微信.jpg.jpg](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003747116-1066810588.jpg)
▲长按加微信群或投稿 ![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003748629-754641840.jpg)
▲长按关注公众号
![3D视觉工坊小助理微信.jpg.jpg](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003747116-1066810588.jpg)
▲长按关注公众号
3D视觉从入门到精通知识星球:针对3D视觉领域的知识点汇总、入门进阶学习路线、最新paper分享、疑问解答四个方面进行深耕,更有各类大厂的算法工程人员进行技术指导。与此同时,星球将联合知名企业发布3D视觉相关算法开发岗位以及项目对接信息,打造成集技术与就业为一体的铁杆粉丝聚集区,近2000星球成员为创造更好的AI世界共同进步,知识星球入口:
学习3D视觉核心技术,扫描查看介绍,3天内无条件退款 ![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003758805-1123548176.jpg)
圈里有高质量教程资料、可答疑解惑、助你高效解决问题 觉得有用,麻烦给个赞和在看~ ![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003801161-1715619362.gif)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003758805-1123548176.jpg)
![](https://img2020.cnblogs.com/blog/1461605/202010/1461605-20201024003801161-1715619362.gif)
微信扫一扫
关注该公众号
个人微信公众号:3D视觉工坊
前往“发现”-“看一看”浏览“朋友在看”