Error when Uploading Custom Model Weights

[habib.matar]

Workspace ID: Intellisee
Project Name: fielddata
Version ID: 3
Project Type: Object Detection
OS: Linux
Private Repository

I am having issues when I try to upload the custom model weights for a version I created. I created a version of ~1100 images with a train, validation, and test set. I downloaded the dataset using the command line API.

Then I used Ultralytics and YOLOv11x to finetune a model on the downloaded dataset. Here is the example of the training script:

from ultralytics import YOLO

# Load a model
model = YOLO("yolo11x.pt")  # load a pretrained model (recommended for training)

#print(model)

# Train the model
results = model.train(data="/mnt/raid/habib/fielddata_v1/data.yaml", epochs=100, imgsz=640, device=[0,1,2], batch=18)

After this completed, I deployed the model to the same version downloaded from Roboflow.

import roboflow

rf = roboflow.Roboflow(api_key="xxx")

project = rf.workspace().project("fielddata")

version = project.version("3")

version.deploy("yolov11", "runs/detect/train7")

This script ran successfully and the model shows up in my Roboflow Model page. However, when I try to use it for LabelAssist or when I try to visualize the model, it errors like the screenshot above.

What do I do to resolve this issue?

[leandro_roboflow]

Hello @habib.matar,

First of all, thank you for reaching out.

I am currently investigating your issue.

Could you please let me know which version of Ultralytics was used to train your model, as well as the version used in conjunction with the Roboflow library? It’s important to note that for YOLOv11, Ultralytics version 8.3.0 or higher is required.

Thank you very much in advance.

Leandro

[Emily]

To clarify, YOLOv11 models must be trained on ultralytics<=8.3.40, here’s a link to our docs to learn more about our model upload requirements.