search

jzhangbs / Vis-MVSNet

Visibility-aware Multi-view Stereo Network
MIT License
235 stars 27 forks source link

pretrained model tank and temple dataset #34

Closed l1346792580123 closed 2 years ago

l1346792580123 commented 2 years ago

I have found that the provided model does not perform well on the tank and temple dataset. Could you please provide the pretrained model on tank and temple dataset?

jzhangbs commented 2 years ago

Hi, thanks for the interest. The provided model is exactly the one we used for benchmarking. And the community usually does not fine-tune the model on TnT dataset. They are either trained on DTU, BlendedMVS or other datasets.

l1346792580123 commented 2 years ago

Thanks for your reply.