Skip to content

Result of model.predict has type ClassificationModel, no keypoints. #345

Open
@LinasKo

Description

Tested on several projects:

pip install supervision roboflow
version = rf.workspace("roboflow-jvuqo").project("football-field-detection-f07vi").version(8)
version.download("yolov8")
kp_model = version.model
assert kp_model is not None

first_image_path = os.listdir(f"{dataset.location}/train/images")[0]
first_image_path = f"{dataset.location}/train/images/{first_image_path}"

result = kp_model.predict(first_image_path).json()

# Fails
keypoints = sv.KeyPoints.from_inference(result)

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions