Original subtitles

Okay, then I was showing you how to train your seven on custom objects in this video.

However, in this video I will do training on Google Collab before doing the training.

Make sure you have installed YOLO v seven on Google Collab to begin launch the browser, then go to

the Google collab URL.

After that, open the URL of seven notebook.

Continue to scroll down.

Navigate to the training section.

The initial cell code, which was used to mark Google Drive when the code cell by pressing this button.

After it will appear like this.

Click Connect to Google Drive.

Exclusive Google Drive account.

Scroll down and then click allow.

Wait until the voting process is finished.

Continue to the next cell.

This cell is used to log into my drive on Google Drive.

Run the code cell by pressing this button.

Continue to the next cell.

This cell is used to enter the root folder of your office seven when the code cell by pressing this

button.

In the previous video, the dataset was uploaded to the data folder.

The following code cell is used to view the contents of the data folder.

Use the command, exclamation mark and less data.

Run the cell by pressing this button.

There is a face mask dataset that was previously uploaded.

Next extract or unzip the dataset.

Use the command exclamation mark, unzip data face mask dataset dot zip.

That's the data.

When to sell by pressing this button.

Wait until the extraction is finished.

Next download waits for training.

Use common exclamation mark w gate.

You are out of your office.

Seven weeks.

When the sale by pressing this button.

Where it to be stored in the root folder of YOLO v seven.

Next, we do the training.

Use the following command.

Exclamation mark.

Python trimmed the pi.

In the bedside.

We read it.

You can increase the batch size value if you're using a Google Calabro.

On device zero.

In the data, right?

The data file that was created in the previous video.

That is face must dot Yemen.

At IMG, we write 640.

In the CFG write the configuration file that was created in the previous video.

That is YOLO.

Five seven Face mask, not GMO.

In width.

We use YOLO five seven training as initial weights.

In the name.

We write the all of his seven face mask.

In head Use helps create custom file in the data folder.

In a box.

We write 300.

Once there to start training.

The following is the training process.

Each epoch will calculate the MLP to get the best weights.

If you want to start training before the specified epochs, you can press this button.

Next we can see the training graph using the tensor board.

Use the following command.

Lonely Extensible.

Sensible.

That's the swap deal.

One strain.

Learn to sell by pressing the following button.

That is an example of a training graph.

This is an accuracy graph.

The higher the better.

This is a lost craft.

The smaller, the better.

Furthermore, the training results will be saved on Google Drive.

That is on runs train model name.

The training is weights file is stored in the weights folder.

This is the weights with the highest p value.

Last the p t is the weights of the last epoch.

In the next video, we will explain how to continue training.

See you then.