ScanNet-SV Dataset Preparation

[ceweiw]

When preparing the ScanNet-SV dataset, the directory structure generated after running the python script load_scannet_mv_data_fast.py is as shown in the following image.
scannet-sv
└── scannet_sv_instance_data
├── scene0187_00_0_axis_align_matrix.npy
├── scene0187_00_0_ins_label.npy
├── scene0187_00_0_sem_label.npy
├── scene0187_00_0_sp_label.npy
├── scene0187_00_0_vert.npy
├── scene0187_00_200_axis_align_matrix.npy
├── scene0187_00_200_ins_label.npy
├── scene0187_00_200_sem_label.npy
├── scene0187_00_200_sp_label.npy
├── scene0187_00_200_vert.npy
├── scene0187_00_400_axis_align_matrix.npy
├── scene0187_00_400_ins_label.npy
├── scene0187_00_400_sem_label.npy
├── scene0187_00_400_sp_label.npy
├── scene0187_00_400_vert.npy
├── scene0187_00_600_axis_align_matrix.npy
├── scene0187_00_600_ins_label.npy
├── scene0187_00_600_sem_label.npy
├── scene0187_00_600_sp_label.npy
├── scene0187_00_600_vert.npy
├── scene0187_00_800_axis_align_matrix.npy
├── scene0187_00_800_ins_label.npy
├── scene0187_00_800_sem_label.npy
├── scene0187_00_800_sp_label.npy
├── scene0187_00_800_vert.npy
├── scene0187_00_1000_axis_align_matrix.npy
├── scene0187_00_1000_ins_label.npy
├── scene0187_00_1000_sem_label.npy
├── scene0187_00_1000_sp_label.npy
└── scene0187_00_1000_vert.npy
When running python tools/create_data.py scannet_mv --root-path ./data/scannet-mv --out-dir ./data/scannet-mv --extra-tag scannet_mv, the following error occurred: The required data files are named inconsistently, and it appears that ScanNetSVData processes only one frame of data per scene.
Traceback (most recent call last):
File "tools/create_data.py", line 78, in
scannet_data_prep(
File "tools/create_data.py", line 18, in scannet_data_prep
create_indoor_info_file(
File "/home/cbdes/zx/embodiedsam/ESAM/tools/indoor_converter.py", line 84, in create_indoor_info_file
infos_train = train_dataset.get_infos(
File "/home/cbdes/zx/embodiedsam/ESAM/tools/scannet_data_utils.py", line 485, in get_infos
return list(infos)
File "/home/cbdes/anaconda3/envs/ESAM/lib/python3.8/concurrent/futures/_base.py", line 619, in result_iterator
yield fs.pop().result()
File "/home/cbdes/anaconda3/envs/ESAM/lib/python3.8/concurrent/futures/_base.py", line 437, in result
return self.__get_result()
File "/home/cbdes/anaconda3/envs/ESAM/lib/python3.8/concurrent/futures/_base.py", line 389, in __get_result
raise self._exception
File "/home/cbdes/anaconda3/envs/ESAM/lib/python3.8/concurrent/futures/thread.py", line 57, in run
result = self.fn(*self.args, **self.kwargs)
File "/home/cbdes/zx/embodiedsam/ESAM/tools/scannet_data_utils.py", line 426, in process_single_scene
points = np.load(pts_filename).astype(np.float32)
File "/home/cbdes/anaconda3/envs/ESAM/lib/python3.8/site-packages/numpy/lib/npyio.py", line 405, in load
fid = stack.enter_context(open(os_fspath(file), "rb"))
FileNotFoundError: [Errno 2] No such file or directory: './data/scannet-sv/scannet_sv_instance_data/scene0187_00_vert.npy'

[xuxw98]

Hi,
If you want to prepare SV dataset, you should run scripts with _sv suffix, rather than _mv. For scannet-sv, see readme here.