Create Manual label jobs

In manual labeling, a label spec is assigned to a manual label job, where a human annotator does the labeling. Manual labeling can refine auto-generated labels or create a seed set for specifying auto labelling jobs. In certain niche domains, the manual labeling can be used as the only labeling workflow. This article explains creating a manual label job for a label spec.

Refer to the following video on submitting a manual label job.

Create manual label job

1. In the left navigation pane, navigate to Repo > Datasets.

2. Locate the required Dataset for which you want to create a manual label job, and click Catalog to open the catalog page.
   

3. Click the down arrow head adjacent to the Visualize button on the top-right corner of the catalog page, and select Auto Label.
   

4. In the New Labeling Job screen, enter the Job Name and Job Description.

5. Select Job Type as Manual Labeling Classification or Manual Labeling Object Detection depending on the spec type for which you are creating a labeling job.

6. Select the label spec for which you want to run the manual label job.
   

7. Enable Prelabeling.
   Use this option if you already have images labeled during the auto label job, but you are unsure about the accuracy of the label, and would want to send such images for manual labeling.

8. Click Submit.
   The manual label job is submitted for manual labeling of the assigned label spec. Once the manual labeling is processed, you can find it on the Label Jobs page for review.

9. On the manual labeling jobs page, review the label job results, and accept or reject the results.
   This works similarly to reviewing the results for auto label jobs.

   1. Navigate to Label > Jobs, and locate the manual label job.

   2. Click Visualize to open the label job page.
      The manual label job results appear under Expert Review Stats, based on the classes defined for manual labeling.
      

10. Verify the results for accuracy of the manual label job.

    You can then create ResultSets based on the reviews and export them to catalog.