For the Nvidia Deep Learning Data Synthesizer (NDDS), I use NVDU Utility (nvdu_viz) for visualizing the 3D BBox.
So, when I try to run it on the produced json file by DOPEWriter from Replicator, I get this error:
(isaac-sim) mona@ard-gpu-01:~/.local/share/ov/pkg/isaac_sim-2022.2.0/first_pass_out_dope_headless$ nvdu_viz RenderProduct_Replicator
args: Namespace(dataset_dir='RenderProduct_Replicator', data_annot_dir='', size=[0, 0], object_settings_path=None, camera_settings_path=None, name_filters=['*.png'], fps=10, auto_change=False, export_dir='', auto_export=False, ignore_fixed_transform=False)
name_filters: ['*.png']
Traceback (most recent call last):
File "/home/mona/.local/bin/nvdu_viz", line 8, in <module>
sys.exit(main())
File "/home/mona/.local/lib/python3.10/site-packages/nvdu/tools/test_nvdu_visualizer.py", line 92, in main
camera_intrinsic_settings = CameraIntrinsicSettings.from_json_file(camera_settings_path)
File "/home/mona/.local/lib/python3.10/site-packages/nvdu/core/camera.py", line 86, in from_json_file
with open(json_file_path, 'r') as json_file:
FileNotFoundError: [Errno 2] No such file or directory: 'RenderProduct_Replicator/_camera_settings.json'
My questions are:
- Do you expect
nvdu_viz
to work with the DOPEWriter json outputs? - If the answer to 1 is no, can you please share a snippet for visualizing the 3D BBox of objects?
My folders are:
(isaac-sim) mona@ard-gpu-01:~/.local/share/ov/pkg/isaac_sim-2022.2.0/first_pass_out_dope_headless$ tree -d .
.
โโโ RenderProduct_omni_kit_widget_viewport_ViewportTexture_0
โ โโโ rgb
โโโ RenderProduct_Replicator
โ โโโ rgb
โโโ RenderProduct_Replicator_01
โโโ rgb
6 directories
Inside each folder there is an RGB folder along with a bunch of json files.
NVDU: Dataset_Utilities/nvdu/viz at master ยท NVIDIA/Dataset_Utilities ยท GitHub
NDDS: GitHub - NVIDIA/Dataset_Synthesizer: NVIDIA Deep learning Dataset Synthesizer (NDDS)