Hello, thank you for open-sourcing the source code. I encountered some problems during the learning process and would like to ask you for advice. I saw that in the code, N virtual points are randomly generated in the instance mask, and the index is obtained by the nearest neighbor distance and distance threshold, but the depth value is directly given by the original point cloud. Is it fixed? I don't understand here.
Hello, thank you for open-sourcing the source code. I encountered some problems during the learning process and would like to ask you for advice. I saw that in the code, N virtual points are randomly generated in the instance mask, and the index is obtained by the nearest neighbor distance and distance threshold, but the depth value is directly given by the original point cloud. Is it fixed? I don't understand here.