Visual ChangeNet-Classification model card lacks public dataset / “Download resources” link

[mangesh.damre]

• Hardware: ⟨T4⟩
• Network Type: visual_changenet_classification
• TAO Version (docker_tag): ⟨nvcr.io/nvidia/tao/tao-toolkit-pyt:v5.5.0⟩
• Training spec file: N/A – dataset missing
• Issue:
The notebook visual_changenet_classification.ipynb contains no dataset path.
Every data-download cell is either commented out or points to an empty string,
so I cannot run the tutorial end-to-end or fine-tune the pretrained weights.

• Reproduction steps:

1. Pulled the TAO container ⟨docker_tag above⟩.
2. Cloned the notebook from
   tao_tutorials/notebooks/tao_launcher_starter_kit/visual_changenet/visual_changenet_classification.ipynb.
3. Executed the notebook until the data-ingestion step; it fails because no dataset path exists.
4. Checked the NGC model card
   Visual ChangeNet Classification | NVIDIA NGC
   – only pretrained weights are available, no “Download resources”

• Request:
– Could NVIDIA share a small, license-free sample dataset (aligned “golden + test” PCB image pairs with CSV labels) compatible with this notebook?
– If not, please update the notebook and documentation to state that users must supply their own dataset and provide guidance on converting public PCB sets (e.g., DeepPCB) to TAO’s CSV format.

Thank you!

[Morganh]

Could you please refer to the dataset mentioned in tao_tutorials/notebooks/tao_launcher_starter_kit/visual_changenet/visual_changenet_segmentation.ipynb at main · NVIDIA/tao_tutorials · GitHub or tao_tutorials/notebooks/tao_launcher_starter_kit/visual_changenet/visual_changenet_segmentation_MVTec.ipynb at main · NVIDIA/tao_tutorials · GitHub.
Thanks.

[yingliu]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks.