Original subtitles

Hello and welcome to this video.

I will explain all of these seven training on custom objects.

This time the training will be conducted on Google Collab mixer.

YOLO v seven is installed on Google Collab.

However, before we begin training we must first prepare the annotated dataset and create a configuration

file.

The first thing we will do is prepare the dataset.

However, preparing the dataset cannot be done directly on Google Collab.

In this example, preparing the dataset will be done on Windows.

First, we make a folder in which to save the dataset.

In this case we will make a folder in the D directory to create a new folder.

Right click.

New folder.

In this example, we name it.

That is a collab for datasets.

You can use datasets that you have annotated.

In this video, we will use an annotated face mask dataset.

The face mask dataset can be accessed and downloaded at the following You are in the following URL.

There is a face mask, data set and split data set dot pi that could be used to split the dataset.

Download the following dataset.

Also download split data set dot pie.

Wait until the download is finished.

When you're finished, go to the downloads folder.

Next move these two files to the folder that was previously created.

In this example, the dataset club folder in the directory look like this, then press control X.

Faced by pressing control fee.

Next extract the dataset.

In this example, extract will use tools from Windows 11 to extract right click, then extract all.

The face mask.

Click extra.

Wait until the extraction is finished.

The following is the annotated face mask dataset following that split the dataset into train validation

and test data.

The split results must match the URL of seven folder structure.

The URL of seven folder structure is shown below.

The images folder contains images, while the labels folder contains annotations, each folder contains

train well and test folders.

We have previously downloaded the Python code for dataset splitting, namely split dataset archive.

The split wants a command prompt in this follow by clicking the address, bar and type CMD.

After that press enter.

Make sure you have Python installed before splitting.

Use the following command to do the splitting python split dataset.

Dot py.

That's the strain.

The train argument is used to set the train that the percentage.

We write 80.

That's just validation.

Validation argument is used to set the percentage of data validation we write in this last test.

The test argument is used to set the percentage of test data.

We write ten.

That's that's fodder.

The folder argument specifies the location of the data set before it is split.

In this case, the face must folder.

This does this the this argument specifies the formula in which the split results are stored.

It will be safe in the face mask dataset folder in this example.

Stress internal.

Wait until the splitting process is finished.

When finished, we turn to Windows Explorer.

There is a face mask dataset folder, which is the result of the split.

The dataset will then be compressed to make it easier to upload to Google Drive.

In this example, we will compress using tools from Windows 11 to compress right click then click Compress

to zip file.

Wait until the compression is finished.

Here are the results.

In the next video, we will create a configuration file.

See you then.