search

microsoft / DIF-Net

Deformed Implicit Field: Modeling 3D Shapes with Learned Dense Correspondence CVPR 2021
MIT License
121 stars 19 forks source link

Some questions about preparation #8

Open ChenhLiwnl opened 2 years ago

ChenhLiwnl commented 2 years ago

Thanks for your outstanding paper! However, I encoutered some problems during preparation work, and hope can get your help. 1) It seems that the provided torchmeta requires torch <=1.4.0, however the torch version in environment.yml is 1.5.0. Will there be some problems? 2) I downloaded traing data which contains 100 .mat files per category, however according to /split/train/xxx.txt files, it seems that each category requires 4000 mats to train. You mentioned that you used mesh_to_sdf to generate the whole training set, but I don't understand how. Could you please tell me?

ChenhLiwnl commented 2 years ago

sorry i noticed that 4000 training data should be downloaded from official site of shapenet, but question 1 still remains

ChenhLiwnl commented 2 years ago

https://shapenet.org/download/shapenetcore well now i downloaded shapenet v2 from here but I don't know how to process it... could you please tell me?

Kiddo1024 commented 2 years ago

Thanks for your outstanding paper! However, I encoutered some problems during preparation work, and hope can get your help.

  1. It seems that the provided torchmeta requires torch <=1.4.0, however the torch version in environment.yml is 1.5.0. Will there be some problems?
  2. I downloaded traing data which contains 100 .mat files per category, however according to /split/train/xxx.txt files, it seems that each category requires 4000 mats to train. You mentioned that you used mesh_to_sdf to generate the whole training set, but I don't understand how. Could you please tell me?

Excuse me, have you solved the question1? Maybe we can have a communication with it!