How to get datasets with var,med and max noise?

[QtEngineer]

Hi, I tried to download ABC and Famous datastes and got vesion of original,free,and extra noise . It seems theres only one noisy datasets. Now I want to get datasets(or test datasets) with different levels of noise.Could you please provided them to me？Thanks！Looking forward to your reply.

[ErlerPhilipp]

ABC original is just called var-noise in the paper. It has the same settings as the ABC training set with random noise strength.

[QtEngineer]

Thanks a lot. I checked ABC test again: abc,abc_extra_noisy and abs_noisefree. I am no sure whether one of them is max noise vesion of 0.05L.

[ErlerPhilipp]

abc_extra_noisy is max noise (0.05L)

[Hxinyue]

Hello, I have my own building point cloud dataset in the format *.ply, what should I do about it that can be used for network training @ErlerPhilipp

[ErlerPhilipp]

Hi @Hxinyue, please see this section in the readme: https://github.com/ErlerPhilipp/points2surf#dataset-from-point-clouds-for-reconstruction Let me know if there are any problems.

[Hxinyue]

Hello, I downloaded a building point cloud data set with the file format of *. ply, such as building_ 01.ply is a point cloud file of a single building. What should I do to use it in network training or final generation? Because there may be many things I don't understand when I first came into contact with this aspect. I look forward to your reply and answer.

------------------ 原始邮件 ------------------ 发件人: "Philipp @.>; 发送时间: 2023年2月4日(星期六) 晚上6:43 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [ErlerPhilipp/points2surf] How to get datasets with var,med and max noise? (Issue #26)

你好@心悦, 请参阅自述文件中的这一部分：https://github.com/ErlerPhilipp/points2surf#dataset-from-point-clouds-for-reconstruction 如果有什么问题告诉我。

— 直接回复此邮件，在GitHub上查看或取消订阅. 你收到这个是因为你被提到了。消息ID：<ErlerPhilipp/观点2 @.***与>.

[ErlerPhilipp]

Training is only possible if you can provide ground-truth meshes. We need the meshes to calculate ground-truth signed distances for supervised training. Assuming you mean reconstruction with final generation, please see my answer in your new issue. You need to choose a pre-trained model (e.g. max-model), create a new point cloud dataset and run a modified evaluation script.

[QtEngineer]

Thanks a lot ！