Xiaoxiao LONG

> cannot import name 'StereoNet'，is that normal? There is no StereoNet model in the code. The code is not complete.

> Thanks for your reply! > > Here is an example of my problem. The second image is scene0000_00/30.jpg, the third image is scene0000_00/60.jpg and the fourth image is the...

> Hi, we address the standard setting of SfM from multiple views, where methods reconstruct 3D up to a global scale. It is not possible to recover global scale using...

Hello, thanks for your interest in our work. Since our model is a learning based method, if your testing data is very different from the training DTU dataset , I...

Hello. Since the sparseneus is only trained on the DTU, which is relatively small dataset and may have domain gap with your data. To prepare a custom dataset, you should...

Yes, we notice that training our model on V100 is very slow, we guess that maybe the sparseconv operations not be optimized well on V100, so we adopt RTX2080Ti for...

You can download the masks here: https://connecthkuhk-my.sharepoint.com/:u:/g/personal/xxlong_connect_hku_hk/EW_v7RA73HNEquScVtNJ34gB4hYlRfEatW4TOg086F0_Lg?e=3SKiif Training without the masks will be not a problem, just simply set the masks to 1.

Thanks for your interests. 1. If use HRNET-48, effective minibatch size 8 across 4 GPUs. 2. We don't freeze any module. 3. Unroll them over an additional dimension in your...

1. We use 3 channels for better visualization. You can adopt more channels here, while in our experiences, 3 channels are enough. The sigma of Gaussian kernel is a hyperparameter,...

Hello. I plan to release the codes in the near future.