All language subtitles for 002 Understanding YOLO dataset annotation format

In this video we will explain about dataset annotation because YOLO is a supervised learning object

detection model, you must use annotated data to train the model.

The annotation provides a bonding box for each object in the image along with the object class name.

This is the annotation format on YOLO.

The first of these values is object class, its value ranging from zero to total class minus one.

The second and third values represent the bounding boxes midpoint.

The midpoint value is relative to the width and height of the image.

X relative to inmates with.

Y relative to image height.

The width of the bounding box is the fourth value.

This width is relative to the image.

With the fifth value is the bonding box is high.

This height is relative to the image height.

Here's an example.

Consider the following example.

There is an image that is 512 pixels wide and 366 pixels high.

In this case, zero represents a mask.

This is the burning boxes midpoint.

X divided by image with.

Equal to this, and it is written as the second value in the annotation.

Why?

Divided by the image.

Equal to this, and it is written as the third value in the annotation.

The width of the box divided by image with.

Equal to this.

And it is is the fourth value in the annotation.

The height of the box divided by image height.

Equal to this, and it is written as the fifth value in the annotation.

There are also several tools available for any city.

YOLO, Mark.

Labor emits.

And let me.

In this course, we will use the label image tools.

The next video will demonstrate how to use label image.

See you then.