Upload of yolov8 weights

Community Help
1

I’m trying the yolov8 object detection flow, using ubuntu and roboflow 0.2.29 and ultralytics 8.0.26 I just got with pip. I create my dataset on roboflow and train locally just fine. I am then falling down on trying to upload the resulting weights to roboflow.
When I run

project.version(DATASET_VERSION).deploy(model_type="yolov8", model_path=f"{HOME}/runs/detect/train/")

I get the error “An error occured when getting the model upload URL: This version already has a trained model. Please generate and train a new version in order to upload model to Roboflow.”

Even though there was nothing for that version in the deploy tab on the roboflow UI prior to running the command. However after running the command even though it says an eror occured the deploy tab now says

" Model upload is processing.
This page will automatically update when it’s ready to use."

with a spinner. But it seems to stay like this forever (at least it has remained in this state for more than 24 hours now).

Any ideas what I might be doing wrong here? Thank you!

2

Hi Mark!

I see you have one Version with a model trained with Roboflow and multiple versions without a model. When you are uploading, make sure you are pointing to one of the generated versions without a model. Give that a try and let me know how it goes.

3

I also have gotten this error with many of my projects. None had models previously. However it seemed to work for some of my models so I’m not really sure what the issue is.

4

@UberSuperBoss @Mark_Blake Hi - we pushed a fix that should resolve this. Apologies for the issue here.

You’ll need to regenerate the version, train, and upload weights to it.

If you have saved the weights, you won’t have to retrain, but you’ll still need to generate a new untrained version to upload the weights to.

5

An error occured when getting the model upload URL: 404 Client Error: Not Found for url:
I just executed all the cells given by default in the colab notebook, and I’m getting this error.