Open Tomoki-0526 opened 3 months ago
Hi, thanks for your interest in our work. For large-scale data, you might want to split it into small patches and run the inference code in parallel. However, the current model has not been trained on noise data. If the data contains serious noise, it is possible to predict an incomplete voronoi diagram and affect the subsequent post-processing pipeline.
Hello, I wonder if your work is valid on a scanned architectual cloud. I run "generate_test_voronoi_data.py" file to generate the mesh but find the output very strange. Here is the origin cloud: but the output is like: So what can i do to preprocess the data?