search

PRBonn / semantic_suma

SuMa++: Efficient LiDAR-based Semantic SLAM (Chen et al IROS 2019)
MIT License
907 stars 205 forks source link

How to install opengl >= 4.0? #24

Closed tym2103 closed 4 years ago

tym2103 commented 4 years ago

Ubuntu 16.04 default opengl version is 3.3. How to update? and when I finish install NVIDIA driver. I can't get opengl info root@ubuntu:~# glxinfo | grep "version" server glx version string: 1.4 client glx version string: 1.4 GLX version: 1.4 OpenGL version string: 1.4 (2.1 Mesa 10.5.4)

Chen-Xieyuanli commented 4 years ago

Hey @tym2103, to install opengl in Ubuntu you could find the information on their official website here.

tym2103 commented 4 years ago

thank for @Chen-Xieyuanli you reply. I follow the step,but the official way also let me download same NVIDA driver. I think I can only read the code without debug. and by the way, can you tell me where the follow formula use in the code

Snipaste_2020-06-21_09-52-41

Chen-Xieyuanli commented 4 years ago

thank for @Chen-Xieyuanli you reply. I follow the step,but the official way also let me download same NVIDA driver. I think I can only read the code without debug. and by the way, can you tell me where the follow formula use in the code

Snipaste_2020-06-21_09-52-41

Hey @tym2103, you could find the semantic icp part in the shader file Frame2Model_jacobians.geom

tym2103 commented 4 years ago

thank for @Chen-Xieyuanli you reply. does SM represent t-1 timestamp semantic map？ and what dose P(u|yu) mean?

Chen-Xieyuanli commented 4 years ago

thank for @Chen-Xieyuanli you reply. does SM represent t-1 timestamp semantic map？ and what dose P(u|yu) mean?

SM is a rendered semantic representation of the Model which includes information not only from the last frame but also from history frames.

We didn't use P(u|yu) in this work. I guess you may ask about P(yu|u), which means the probability of the semantic class giving a vertex.

I hope this helps!

tym2103 commented 4 years ago

@Chen-Xieyuanli 感谢大佬的耐心回复，下面的问题我不太会用英文描述，看到你的CSDN博客知道你懂中文，我就直接用中文说了，^_^。 SD的观察视角是当前的局部坐标，SM的观察视角是世界坐标（也就是起点的坐标）还是t-1或者t时刻的局部坐标？ 然后下面P(yu|u)我说一下我的理解，假设语义地图u的大小为100200，然后车这一类占了10个点，那么P(yu|u)的大小是不是就为10/(100200),这样理解对吗？

Chen-Xieyuanli commented 4 years ago

@Chen-Xieyuanli 感谢大佬的耐心回复，下面的问题我不太会用英文描述，看到你的CSDN博客知道你懂中文，我就直接用中文说了，^_^。 SD的观察视角是当前的局部坐标，SM的观察视角是世界坐标（也就是起点的坐标）还是t-1或者t时刻的局部坐标？ 然后下面P(yu|u)我说一下我的理解，假设语义地图u的大小为100_200，然后车这一类占了10个点，那么P(yu|u)的大小是不是就为10/(100_200),这样理解对吗？

No problem! We can have a further discussion via e-mail in Chinese. Here is my email address: xieyuanli.chen@igg.uni-bonn.de

Since the discussion is more about the details of SuMa++ paper which is out of the scope of this issue, I am going to close it and will reply to you via email.

Thank you for following our work and look forward to your email!