generalization ability（泛化能力） of neural network based MVS about cvp-mvsnet HOT 2 OPEN

Hi Jiakui,

Yes you can do that.
The model we reported in the paper is trained on DTU dataset.
That model can achieve reasonable reconstruction quality on the Tanks and Temples dataset, which contains real-life scenes different from the toys in DTU dataset.

Cheers,
Jiayu.

from cvp-mvsnet.

嗨家葵，

是的，你可以这么做。 我们在论文中报告的模型是在 DTU 数据集上训练的。 该模型可以在 Tanks and Temples 数据集上实现合理的重建质量，其中包含与 DTU 数据集中的玩具不同的真实场景。

干杯， 佳玉。

Thank you very much for the author's paper and code, but I found a dimension error after modifying the data set configuration code and path, so I would like to ask if you can send the test code of T&T to my email address? Thank you very much and look forward to your reply: )

from cvp-mvsnet.