Priority scores

Upload priority scores (Active Learning)

The priority scores feature helps you determine the items that you need to annotate first to improve your model training. It's a key active learning feature that speeds up your annotation time.

To upload priority scores:

1. Go to Setup → Settings → General.
2. Enable Priority scores. When enabled successfully. you'll see a message at the top of the page that says Changes saved.
3. Go to Data, click Add, and select Upload Priority Scores.
4. Drag and drop or choose a CSV file from your computer.

Who can upload priority scores?

Team Owners, Team Admins, and Project Admins (with permission) can upload priority scores.

❗️

• Items whose names don’t match the ones in the CSV file or have no priority scores will be skipped.
• Items that don't have names or priority scores won't be uploaded.

After the upload is complete, a priority score appears next to each item. You can sort your items by their priority values.

If the items have priority scores and the Annotator requests tasks, they’ll receive the items sorted by priority score instead of by item name.

📘

Download this file and use it as a guide to create a CSV file.

SDK function

To upload priority scores:

project_name = 'Project Name'
scores = [
    {
        "name" : "image_1.png",
        "priority": 0.567
    },
    {
        "name" : "image_2.png",
        "priority": 0.789
    }
]

sa.upload_priority_scores(
    project = project_name,
    scores = scores
)