search

XuyangBai / D3Feat

[TensorFlow] Official implementation of CVPR'20 oral paper - D3Feat: Joint Learning of Dense Detection and Description of 3D Local Features https://arxiv.org/abs/2003.03164
MIT License
259 stars 38 forks source link

Aligned or unaligned pts in Training and Testing datasets #15

Closed BingCS closed 4 years ago

BingCS commented 4 years ago

Hi Xuyang,

I just noticed that you saved the aligned point cloud points as the training dataset but the testing dataset from 3DMatch are unaligned during the testing phase. May I confirm this with you? Many thanks.

Best, Bing

XuyangBai commented 4 years ago

Hi @BingCS Exactly, I align the point cloud using gt transformation and add rotation translation augmentation for training (to achieve rotation invariant feature descriptors). But for the testset, I directly download it from 3DMatch for a fair comparison with other methods.

Best, Xuyang

BingCS commented 4 years ago

Thanks for your reply!