Pointnet_Pointnet2_pytorch icon indicating copy to clipboard operation
Pointnet_Pointnet2_pytorch copied to clipboard

se_seg模块,我看了斯坦福数据集,他有六列数据,后三列是rgb信息,请问,我的数据集没有颜色信息,还能用来训练吗?》

Open lzh1998-jansen opened this issue 3 years ago • 9 comments

lzh1998-jansen avatar Oct 20 '21 07:10 lzh1998-jansen

Yes? There's an option "normal_channel" for this, just set it to False, and the network will not expect rgb channels

JohanYe avatar Oct 21 '21 07:10 JohanYe

very thanks to you ,guy, i am labeling datasets which are not standard,it's a little difficult,i think ,i can use semantic segmentation to slove my task.

------------------ 原始邮件 ------------------ 发件人: "yanx27/Pointnet_Pointnet2_pytorch" @.>; 发送时间: 2021年10月21日(星期四) 下午3:24 @.>; @.@.>; 主题: Re: [yanx27/Pointnet_Pointnet2_pytorch] se_seg模块,我看了斯坦福数据集,他有六列数据,后三列是rgb信息,请问,我的数据集没有颜色信息,还能用来训练吗?》 (Issue #134)

Yes? There's an option "normal_channel" for this, just set it to False, and the network will not expect rgb channels

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

lzh1998-jansen avatar Oct 21 '21 07:10 lzh1998-jansen

你好,请教下,你看懂作者给的6维的数据集的含义了吗?出来x,y,z后面三个是什么呀。

hongyiWeng avatar Nov 01 '21 03:11 hongyiWeng

你好,请教下,你看懂作者给的6维的数据集的含义了吗?出来x,y,z后面三个是什么呀。

Your answer is in the title of this issue, lzh1998-lzh says it is RGB colors at each point

JohanYe avatar Nov 01 '21 09:11 JohanYe

very thanks to you ,guy, i am labeling datasets which are not standard,it's a little difficult,i think ,i can use semantic segmentation to slove my task. ------------------ 原始邮件 ------------------ 发件人: "yanx27/Pointnet_Pointnet2_pytorch" @.>; 发送时间: 2021年10月21日(星期四) 下午3:24 @.>; @.@.>; 主题: Re: [yanx27/Pointnet_Pointnet2_pytorch] se_seg模块,我看了斯坦福数据集,他有六列数据,后三列是rgb信息,请问,我的数据集没有颜色信息,还能用来训练吗?》 (Issue #134) Yes? There's an option "normal_channel" for this, just set it to False, and the network will not expect rgb channels — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

Have you finished the semantic segmentation on your own datasets? And how did you labeled datasets and make the same format as the Stanford3dDataset?

bwlstart avatar Feb 09 '22 03:02 bwlstart

@lzh1998-lzh how did you labeled datasets and make the same format as the Stanford3dDataset?thank you

hongzhiqiang67 avatar Mar 19 '22 04:03 hongzhiqiang67

@lzh1998-lzh my .las point cloud data only have (x,y,z,lable),How can i make the same format as the Stanford3dDataset? Can I add a QQ to learn from you。thank you!!!! my qq:1002923065

A-Chas avatar Jun 08 '22 03:06 A-Chas

@lzh1998-lzh my .las point cloud data only have (x,y,z,lable),How can i make the same format as the Stanford3dDataset? Can I add a QQ to learn from you。thank you!!!! my qq:1002923065

There's an option "normal_channel" for this, just set it to False, and the network will not expect rgb channels。 没有颜色 rgb 信息也可以。

lzh1998-jansen avatar Jun 15 '22 11:06 lzh1998-jansen

Hello where is the "normal_channel" ?

qq474477759 avatar Oct 11 '22 08:10 qq474477759

it's set by default to false, for part segmentation (pointnet2_part_seg_ssg.py) classifier = MODEL.get_model(NUM_CLASSES,normal_channel=False)

anassmu avatar Dec 20 '22 13:12 anassmu