Unable to upload model

[Jay_Fischer]

I’ve been trying to upload a model to my project from a local Jupiter notebook. It kept coming back with an error saying that the model failed to upload. Took a look at forum and saw that images must be square. So I created a version with 640X640 dataset and added the imgsz=640 argument. Yet still, it will not upload and keeps telling me the upload failed.

[Mohamed]

HI @Jay_Fischer - can you try this process:

1. Generate a new version of the dataset

2. To export the dataset, and begin model training, click the appropriate model architecture here:
   

3. Select “Get Snippet”
   

4. Click the small “copy” icon:
   

5. Open the appropriate notebook from the notebooks Repository, and “Save a Copy in Drive”

6. Follow the training process outlined in the notebook, and be sure you have explicitly set the img or imgsz parameter in the training code, based on how you’ve resized your images – Preprocessing, Resize Documentation:
   
   

   

• Example for YOLOv5 object detection
  

  

• Example for YOLOv7 Instance Segmentation
  

  

• Example for YOLOv8 object detection
  

  

7. Once the training has completed, proceed to the version.deploy() step, where you will run the function to upload your model weights

---

If you see an error again, after going through this process, please reply to me here.

[Jay_Fischer]

@Mohamed redid the model and it failed again. Project is FTC GEN 8872 CLASSIFIED Object Detection Dataset (v9, Please Work) by Test

[Jay_Fischer]

[Jay_Fischer]