WEBVTT

00:00.080 --> 00:01.000
Welcome back.

00:01.040 --> 00:08.120
We finished training our models and we get the hard disk and the airports.

00:08.160 --> 00:12.480
We created the two classes hard disk and airports.

00:12.520 --> 00:14.880
Now let's train the model.

00:14.920 --> 00:17.240
You can click Advanced Apex.

00:17.280 --> 00:17.760
Apex.

00:17.760 --> 00:26.320
As I told you, one epoch means that each and every sample in the training dataset has been fed through

00:26.320 --> 00:28.880
the training model at least once.

00:29.000 --> 00:36.360
If your apex are set to 50, for example, it means that the model you are training will work through

00:36.360 --> 00:39.680
the entire training data set 50 times.

00:39.840 --> 00:46.280
Generally, the larger the number, the better your model will learn and predict data.

00:46.320 --> 00:51.520
As we did in the previous sections, I'm going to use 100.

00:51.560 --> 00:57.680
The batch size a batch is a set of samples used in one iteration of training.

00:57.840 --> 01:00.720
Keep it 16 learning rate.

01:00.760 --> 01:03.080
Be careful of tweaking this number.

01:03.120 --> 01:09.440
Even small differences can have huge effects on how well your model learns.

01:09.560 --> 01:15.480
Okay, click train the model and you see preparing training data.

01:15.520 --> 01:16.880
You get this note.

01:16.920 --> 01:20.600
You must leave this tab open to train your model.

01:20.600 --> 01:21.440
Click okay.

01:21.760 --> 01:28.950
And here we are getting training progress in the next sections, we're going to learn how to train the

01:28.950 --> 01:36.590
models, how to detect objects and and work with images using the Google Colab.

01:36.590 --> 01:42.750
And we'll go into create our custom model in the same way, but without using teachable machine, we're

01:42.750 --> 01:44.590
going to use our code.

01:44.590 --> 01:52.470
So we're going to create codes as we did in the previous sections, but we're going to use it with the

01:52.750 --> 01:55.510
images for object detection models.

01:55.630 --> 01:58.470
Now we have to export your model.

01:58.470 --> 02:01.430
We have the input a webcam.

02:01.430 --> 02:06.070
So go and rerun the webcam.

02:06.230 --> 02:13.110
And we see guys that in this preview we have this is the AirPods is 100.

02:13.150 --> 02:17.950
This is the hard disk is 100 and it's well trained.

02:17.950 --> 02:21.150
So change the position with the keyboard.

02:21.270 --> 02:29.230
You get the hard disk and you see guys the percentage of the hard disk increases and you see guys that

02:29.230 --> 02:32.950
the hard disk is well detected in our model.

02:32.950 --> 02:34.350
Let me go to the AirPods.

02:34.350 --> 02:37.470
You see guys how fast is detection.

02:37.470 --> 02:46.590
Because we will train our model with hundreds of images and we make 100 epochs for each class.