Use of annotation tool


This page uses ABEJA Platform GUI and Annotation Tool to create a dataset.

Dataset creation using annotation tool

Use ABEJA Platform’s Annotation Tool to annotate the stored images and create a dataset.

Creating an annotation project

Click “Annotation” in the left menu. Click “Authorize” to go to another screen.

Next, click “Create a new project” from the upper right of the transitioned screen.

Select “Image Classification” from “Create a new Project” and select “Next”.

Next, select “Annotation by select” and enter the following items in the Category 1 and Labels fields.

Category 1

id Name
0 clothes


id Name
0 jeans
1 denim_skirt

After entering, select “Next”.

Item Value Remark
Name Any Value Project Name
Number of data items Any Value -
Results of each data item 1 e.g : Classification
Targets number - -
Input datalake channel Select the channel of the created datalake -

Confirm the last entry and select “Save”.

Annotation work

Next, work on the created annotation project.

Click the work screen from the gear icon in the upper right of the created project.

The category set on the right side of the screen is displayed. Select “Send” at the bottom of the screen.

Output the results as a dataset

Annotation results can be output as an ABEJA Platform dataset. It can be used as learning data as it is.

Go to the management summary page.

Click the gear icon on the upper right of the created project and click “To worker page”.

Item Value
Dataset Name Any dataset name

Click “OK” and you will see a success message.

Check the created dataset

The exported dataset can be confirmed in the management console of ABEJA Platform.

Open the management console of ABEJA Platform and click “Training> Datasets” in the left menu. You can select the created dataset and check the “Preview” of the dataset.

You can also display only images with the corresponding label from Labels on the right side of the screen.

In the next model training, the model is retrained using the dataset created this time.