Jihyeon Yi

For your information, the OTE model I've tried is `efficientnet_b2b-mask_rcnn-480x480` at https://github.com/openvinotoolkit/training_extensions/tree/develop/models/instance_segmentation/model_templates/custom-instance-segmentation

Thank you @druzhkov-paul As I understand, it could be solved if I use or implement the correct annotation loader. Is it correct?

Hi @DP1701, If your task is one of classification, detection, segmentation, or re-id, you can use [task specific splitter](https://github.com/openvinotoolkit/datumaro/blob/eb4435544e410d5e13397e99dff5f4d1ec571cee/src/datumaro/plugins/splitter.py#L28) for your purpose. You can use `Split` class directly or call...

Hi @DP1701 Let me correct the transform call like below. ```python dataset = dataset.transform('split', task='detection', splits=splits) ```

I close this ticket because it seems that your question is resolved.

Hi @DP1701, I'm happy to hear that you successfully split your dataset. For the second question, you've got two json files because the coco format generates separated annotation files for...

Thanks for your interest in datumaro project. We will consider supporting the new format.