"Could not clone features for asset. Try refreshing or proceed without template."

Using Labelbox
1

Hello Labelbox,

I am using MAL this week and learning how to use it.
Today, I updated the detectron2 model with labels and tried to upload it to my project.
MAL looks appropriately uploaded, but when I get back to labeling, I see the error like the title.
It not only cannot load MAL pre-annotation, but also it seems it has trouble loading data rows properly.
The error still happens with MAL disabled and deleted.

Is there any suggestion to deal with this?

Screen Shot 2022-07-01 at 12.18.08 PM
Screen Shot 2022-07-01 at 12.18.08 PM1920×1130 132 KB

I changed external ids in the dataset before training the new model.
I wonder change the external ids caused these whole errors.

2

Is this still an issue for you?

3

Yes, it is still happening and still an issue.

4

Can you share the annotation mask example?
(It might be an issue with the annotation mask, perhaps the color specified is incorrect or the shape is too complex)

5

Hello @manu,

Thanks for checking in. Here is one of the raw data.

[{"label":"ID","value":"cl4yuh2g0007e078chg2pheoi"},{"label":"External ID","value":"100000102622.jpg"},{"label":"Created","value":"15 days ago"},{"label":"Asset type","value":"Image"},{"label":"Dataset","value":"limes","link":"/data/cl47mvicn0mvu077sf68q4uvm"},{"label":"Row data","value":"https://storage.labelbox.com/cl47kik8h2a1z07b43lbl9url%2Ff4eea7fe-838e-b97d-ba28-ac515d07a0ed-100000102622.jpg?Expires=1657832245314&KeyName=labelbox-assets-key-3&Signature=YrQvzpJITYgiItOOb97FyVz4u18","link":"https://storage.labelbox.com/cl47kik8h2a1z07b43lbl9url%2Ff4eea7fe-838e-b97d-ba28-ac515d07a0ed-100000102622.jpg?Expires=1657832245314&KeyName=labelbox-assets-key-3&Signature=YrQvzpJITYgiItOOb97FyVz4u18"}]
6

Thanks. For importing model prediction, you must use a segmentation mask. Here’s a google colab for reference: Google Colab

(more info here: Model-assisted labeling (MAL))

7

hello @manu,

It’s quite confusing to understand your reply. I am not importing model prediction here. the project cannot load its asset.

8

Hi @GuuintheHaus, in the screenshot above, the error is caused because the model prediction (using model-assisted labeling workflow) on that data row is unable to be converted to a pre-label. We convert model predictions to pre-label at the time of labeling session.

If you are able to reproduce this issue, i recommend that you reach out to Labelbox support so that they can gather logs for additional debugging. Customer support