Remove data rows from project

Using Labelbox Annotate
1

I accidently added data rows into an annotation project. What is the best way to remove them from the project? I don’t want to delete the data itself, just the connection to the specific project.

2

Hi @liorr ,

From the Project → Data rows tab go to Batches, use the ellipsis on the right and :

Labelbox 2023-09-11 11-39-11(1)(1)

You can also delete the batch entirely (if no labels have been created).

Many thanks,
PT

1 Like
3

Thanks! But that removes the entire batch, right? What if I just want to remove parts of the batch?

Lior

1 Like
4

At the moment there is no way to remove individual data rows at the batch label, we advocate to delete the batch (or remove the asset queued) and re-create a batch.
We are also evaluating the possibility in the future do more cherry picking around batches.

Many thanks,
PT
Labelbox Support

3 Likes
5

@ptancre I would really love to be able to delete at the data row level instead of the batch level. It’s a big pain point for me when I am pulling in large sets of data for review and then finding things that don’t belong and can’t confidently delete it without interfering with the rest of the data from the batch.

6

Our process is designed to handle the curation in the Catalog to make sure your pipeline gets only the relevant data and nothing else. This way, we can cut out any noise and save time and money during labeling and reviewing.

We added a batching system for maximum flexibility, but once we set up the queues and reservations, it becomes tricky to change things on the fly.

My earlier answer still stands, but if you’ve got labeled or skipped data, you could ask your reviewers to add them to the rework task. Then, just delete those labels in bulk and remove the unnecessary data rows from the queue. Do this towards the end of your project to prevent any hiccups in the workflow or looping irrelevant data rows.

So, in a nutshell, spending more time on curation upfront in the Catalog is super helpful. That’s why we’re still working on adding new features to make this process faster and smoother for you.

1 Like
7

I understand - and sounds solid. It still stands that from a user perspective that once we need to delete a data row from a project it feels overly complicated. Both things can be true, that there are valid reasons for the pain point to exist and for the user pain point to exist.

1 Like