tangjiapeng
tangjiapeng
Before running the demo, you need to create an environment and compile the extension modules required by the README.md under the "Skeleton_inference/Volume_refinement/Mesh_refinement" folder.
Sorry, I find that there is a small error in the L23 in ply2mat.py. Please replace this line "if not os.path.exists('./ShapeNetPointCloud' + cat):" with "if not os.path.exists('./ShapeNetPointCloud/' + cat):", and...
Thanks for your feedback! I will re-publish it again.
Hi, if you want to reconstruct colored surface meshes, you can extend our model to learn color fields that are defined at the near surface region. Give a point sampled...
Hi, The GPU RAM when I run these experiments was at least 11 GB. If you are using 8GB GPUs, you can reduce the batchsize from 16 to 12 or...
The batch_size in demo_matterport.yaml was 2, you can set it to 1. A better choice is to use GPUs with larger RAM.
Hi, xinrui, have you addressed the memory issue?
After downloading the demo data, you need to use the preprocessing script to prepare the cropped scenes for test-time optimization. And then you can run the test-time optimization. python scripts/dataset_matterport/make_cropscene_dataset.py...
I have fixed this bug by providing a new script (scripts/dataset_matterport/make_cropscene_dataset2.py) for processing Matterport3D demo scenes. Please refer to the README.md.
Hi, thanks a lot for your interest. The way to call the multiple GPUs to run the test-time optimization is similar to what we usually do in the multi-GPUs training...