search

suyz526 / ZebraPose

The implementation of the paper 'ZebraPose: Coarse to Fine Surface Encoding for 6DoF Object Pose Estimation' (CVPR2022)
MIT License
122 stars 20 forks source link

Pretrained LM-O weights #27

Closed LeroyChou closed 1 year ago

LeroyChou commented 1 year ago

Hi, Thank you for your great work. I'm very interested in your proposed method and wanna reproduce the results on LM-O, but it seems that you didn't provide pretrained weight of each object. It would be very kind if you make the pretrained weights publicly available. Best regards Leroy

suyz526 commented 1 year ago

Hi Leroy,

thanks for your interested.

We have provided model for each object. In the folder paper, we provided the model trained with real+pbr images. In the folder bop, the LM-O objects are trained only with pbr images (as the BOP challenge requires). For the symmetric objects that are symmetry-aware trained, their weights are saved in bop_sat. The bop_sat folder has less models, since the missing models can be taken from the bop folder.

Please also note that the LM-O dataset has less objects than LM dataset.

Best Regards, Yongzhi

LeroyChou commented 1 year ago

Oh yes, I found it. Thank you so much!