Appendix: Types of Templates

Here is a list of templates available in the Annotation Tool.

目次

Object Detection
Semantic Segmentation
ImageClassification
TextClassification
TextExtraction
Image to Text
Polygon
Polyline
Private Templates

1. Object Detection

You can perform the task of detecting objects in an image.
You can add bounding boxes to an image.
Examples: product detection, object counting, face detection
Supported formats: jpg,png
Output: json

2. Semantic Segmentation

You can perform the task of extracting an area of an image.
Color in the area to be extracted for an image.
Performs the task of extracting the object shape in the image.
You can draw a polygon by hitting a vertex on the object to be detected in the image and color in the area.
Examples: Area Extraction, Area Extraction, and Flaw Detection
Supported formats: jpg,png
Output: json,png

3. Image Classification

You can perform the task of image classification.
You can add pre-set tags to images.
Example: Adding product classification, emotion classification and attribute information
Supported formats: jpg,png
Output: json

4. Text Classification

It enables you to classify texts. You can assign preset tags in body text.
Example: text category classification, toxic content classification, text emotion classification
Supported formats: text (UTF-8)
Output: json

5. Text Extraction

It enables you to assign a preset tag on the target word in body text for extraction and classification.
Example: word extraction, word classification
Supported formats: text (UTF-8)
Output: json

6. Image to Text

It enables you to extract text information from an image.
You can assign free text information on images.
Example: text recognition (OCR), assigning metadata to images
Supported formats: jpg,png
Output: json

7. Polygon

It enables you to extract the shape of an object in an image.
It enables you to draw polygons by joining adjacent vertices on your target object on an image and extract vertex coordinates.
Example: object contour extraction, text detection
Supported formats: jpg,png
Output: json

8. Polyline

It enables to trac lines from images and pinpointing one pixel.
It enable to draw points or lines while striking the vertices of the object you want to detect in the image, and extract the coordinate information.
Example: posture estimation, people counting
Supported formats: jpg,png
Output: json

4. Private Templates

You can use an existing template or create a private template customized to your requirements.

Creating a template

You can create your own template.
Please refer to the Developer’s SDK document here.

Request a template to be created

ABEJA can also create a template for you.
Please use the chat box in the bottom right corner of the ABEJA Platform console to request a template.

Updated on 19 Aug 2020