Skip to main content

Auto Labeling - Object Detection

Description​

Auto labeling is a tool that allows the user to automatically label objects in images for Object Detection tasks from the results of the selected tools. This is useful for labeling large amounts of images.

caution

The accuracy of the auto labeling depends on the accuracy of the results from the selected tool. Therefore, the label files should be checked for before using them for training.

Settings​

Tool​

A tool to reference the input bounding box to auto labeler. The bounding boxes of the selected tool will be used to label the objects in the image.

The following tools are available:

caution

In order to use the Auto Labeling tool, the user must first run the reference tool.

Class Name​

Class name to reference the detected box. The class name will be used to label the objects in the image. For some tools, the class name is taken directly from the tool.

These tools are:

Export Directory​

Path to the directory where the image and the label files will be exported.

Image File Format​

The format of the exported image file.

The following formats are available:

  • JPEG
  • JPG
  • PNG
  • TIFF

Label Type​

The type of the exported label file.

The following types are available:

Pascal VOC​

The Pascal VOC format is an XML file that contains the bounding boxes of the objects in the image.

Here is an example of a Pascal VOC label file:

<annotation>
    <folder>images</folder>
    <filename>image.jpg</filename>
    <path>C:\Users\user\Documents\images\image.jpg</path>
    <source>
        <database>Unknown</database>
    </source>
    <size>
        <width>1920</width>
        <height>1080</height>
        <depth>3</depth>
    </size>
    <segmented>0</segmented>
    <object>
        <name>person</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>0</xmin>
            <ymin>0</ymin>
            <xmax>100</xmax>
            <ymax>100</ymax>
        </bndbox>
    </object>
    <object>
        <name>person</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>100</xmin>
            <ymin>100</ymin>
            <xmax>200</xmax>
            <ymax>200</ymax>
        </bndbox>
    </object>
</annotation>

YOLO​

The YOLO format is a text file that contains the bounding boxes of the objects in the image.

Here is an example of a YOLO label file:

0 0.5 0.5 0.1 0.1
1 0.5 0.5 0.1 0.1

The class name is not included in the label file, but it is included in the class names file.

person
dog

Where the first line of the class names file corresponds to the class with the index 0 (person), the second line corresponds to the class with the index 1 (dog), and so on.

info

Note that x_center y_center width height are normalized between 0 and 1, and the format is class x_center y_center width height.

Image Compression​

Export image compression level (0%: No compression - 100%: Max compression). The default value is 5%.