search

liuzhengzhe / DreamStone-ISS

This is the project for DreamStone: TPAMI & ISS: ICLR 2023 spotlight
https://liuzhengzhe.github.io/DreamStone.github.io/
43 stars 3 forks source link

Evaluation issues regarding training phase 1 #7

Open happysxpp opened 10 months ago

happysxpp commented 10 months ago

1.May I ask how to evaluate the training results of Phase 1?

2.May I ask how mIOU is obtained in Figure 4 (b)?

Thank you very much, if you could reply to me!

liuzhengzhe commented 10 months ago

Hi,

  1. FID: we adopt this implementation. https://github.com/mseitzer/pytorch-fid FPD: https://drive.google.com/file/d/1vniFpLFZwDfwMT3Ce2KXNQB8Bdv65iUG/view

  2. Generate all the shape in 128^3 voxelized volume of the test set, and then evaluate the 3D IoU between reconstructed shape and the GT. 3D_IoU=intersection of the occupied voxels/union of the occupied Voxels

Zhengzhe

⑨ @.***> 于2023年10月31日周二 23:20写道：

1.May I ask how to evaluate the training results of Phase 1?

2.May I ask how mIOU is obtained in Figure 4 (b)?

Thank you very much, if you could reply to me!

— Reply to this email directly, view it on GitHub https://github.com/liuzhengzhe/DreamStone-ISS/issues/7, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB7WPDCZZJNV7ZDWNOZIMALYCEJMFAVCNFSM6AAAAAA6X4EEDOVHI2DSMVQWIX3LMV43ASLTON2WKOZRHE3TANZRGA3TGMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>