Open cama36 opened 8 months ago
I am not entirely sure of your question!
If you have many small processing units, the best way to link these together is by using a tile index. This is a text space delimitated file with the columns [tile_name x y z file_path] for each tile/small processing unit. Both semantic.py
and instance.py
can use this to identify neighbouring tiles.
If you can provide more detail the hopefully I can give a clearer answer.
Hello, thank you very much for your excellent work. However, I have a question. First, we will process the entire scene into processing units one by one, and handle instance segmentation on each unit. However, after processing each small point cloud, how can we merge these small blocks to solve the inconsistency of the results and obtain the unique result of the entire scene?