Rui Peng
Rui Peng
The fusibile script has insufficient read or write permissions in a directory. You can generate the depth map and confidence map first, and then run the fusibile command with 'sudo'.
We determined the parameters of each scene during the experiment of reproducing our baseline.
You can find them in function [unified_focal_loss](https://github.com/prstrive/UniMVSNet/blob/7c1c9862cc6f2eb0b74c0569d58826d9416195b5/loss.py#L61).
We used two V100 GPUs.
We set the conf values according to our experiments on our baseline model. And it should help if you have a robust network or design a more adaptive strategy like...
In our experiments, we use the camera setting of MVSNet and CasMVSNet. I guess the unsatisfactory results on your custom data may lie on the unsatisfactory camera parameters. Maybe you...
The temperature is an attribute of class Routing. Therefore, its value can be modified in each epoch. What you need to do is to use the Tensorflow API to find...
Yea, this is true. But this is a mechanism limited by Tensorflow.
Why is it that after running the test, the point cloud is divergent and the depth map is also blurry
Maybe there are something wrong on the setting of your camera parameters.