Pose Estimation Dataset Issues

Community Help
1

Hi,
I used roboflow to create a dataset for horse pose estimation but I have an issue when I try to use it.
I Downloaded the dataset with the Yolov8 format for a Yolov8 model but when I run it I get this error message :

labels[“keypoints”] = torch.from_numpy(instances.keypoints)
TypeError: expected np.ndarray (got NoneType)

(This is only the last few lines of the long error message)
The code is only 3 lines :

from ultralytics import YOLO

model = YOLO(“yolov8n-pose.pt”)
results = model.train(data=“horse_posev8.yaml”, epochs=100, imgsz=640)

I hope anyone can explain me why it doesn’t work and how to fix it !
Thanks in advance !

2

Hi! Can you check the docs and confirm you’re using the right version of ultralytics?

3

Actually, can you verify if training a model within roboflow works on that dataset? That should indicate whether or not it’s a dataset / roboflow error

4

I was on the wrong version, didn’t think the latest version couldn’t handle previous features,
Thanks !

5

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.