TAO Toolkit

• Hardware (RTX 3080 Ti)
• Network Type (Mask_rcnn)
• TAO Version
(Configuration of the TAO Toolkit Instance
dockers: [‘nvidia/tao/tao-toolkit-tf’, ‘nvidia/tao/tao-toolkit-pyt’, ‘nvidia/tao/tao-toolkit-lm’]
format_version: 2.0
toolkit_version: 3.22.05
published_date: 05/25/2022)

Hello everyone!

I want to train a custom dataset on the MaskRCNN model in the TAO Toolkit. The annotations in the example dataset are in json format and there are a lot of json files in the downloaded annotations folder, instead of just one each for training, testing and validation (as is the standard), which is making it really hard to make sense of how to fit my own dataset into the notebook. I am using Roboflow to generate json coco files, which I want to reference in the specs file.

This is the file system of the Roboflow annotations:

  • train

    • ‘_annotations.coco.json’
    • [jpg training images]
  • test

    • ‘_annotations.coco.json’
    • [jpg test images]
  • valid

    • ‘_annotations.coco.json’
    • [jpg valid images]

Do you guys know how I can fit this dataset into the TAO MaskRCNN training pipeline?

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.

Did you ever try to run the Mask_rcnn jupyter notebook? I think it can help you understand the whole steps.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.