WEBVTT

00:00.080 --> 00:00.840
Welcome back.

00:00.840 --> 00:01.760
Congratulations.

00:01.760 --> 00:06.560
We created our model inside this folder called runs.

00:06.560 --> 00:09.040
So open runs detect train.

00:09.040 --> 00:11.760
And here we have the results.

00:11.920 --> 00:15.720
Also we have a folder called weights best and last.

00:15.720 --> 00:18.600
This is our best trained model.

00:19.320 --> 00:23.320
Now let's add the validate our model.

00:23.320 --> 00:26.520
So in this video we're gonna validate the model.

00:26.520 --> 00:32.720
Create a new code here from ultra import YOLO.

00:32.920 --> 00:35.880
And we need to load the model.

00:35.880 --> 00:36.920
What model.

00:36.920 --> 00:41.560
We need to load this model that created one.

00:41.560 --> 00:50.720
So copy the path of this best model and paste it here inside this cell okay.

00:50.920 --> 00:51.880
Like this.

00:52.360 --> 00:54.680
And let's let's run.

00:54.720 --> 00:55.440
Here we go.

00:55.480 --> 00:58.440
It's running safe with no errors.

00:58.480 --> 01:00.920
Now let's validate the model.

01:00.920 --> 01:03.400
So validate the model.

01:03.440 --> 01:04.320
How to validate.

01:04.320 --> 01:15.520
We're going to allow the model to use the validate the validate images and use them metrics equals to

01:15.560 --> 01:17.520
model dot val.

01:17.560 --> 01:18.560
Let's run.

01:18.560 --> 01:19.520
And here we go.

01:19.680 --> 01:25.510
We are validating our model results saved to content runs.

01:25.510 --> 01:28.430
So again guys, let me go to the runs.

01:28.670 --> 01:28.950
Here.

01:28.950 --> 01:31.990
We have runs detect validate.

01:31.990 --> 01:35.950
So here is used for training and here used for validation.

01:36.190 --> 01:42.670
Open the val and let me see that the box F1 curve.

01:42.710 --> 01:46.870
You see guys this is the F1 confidence curve.

01:47.030 --> 01:50.950
You can you can open the box PR curve.

01:50.990 --> 01:52.470
This is the second one.

01:52.470 --> 01:53.910
We are reaching one.

01:53.910 --> 01:56.590
And the battery is 0.995.

01:56.630 --> 01:59.670
It's detecting and working fine here.

01:59.670 --> 02:01.630
The precision confidence curve.

02:01.630 --> 02:06.710
It's increased to reach one all classes and the battery.

02:06.870 --> 02:10.710
You see the battery curve box R curve.

02:10.870 --> 02:21.110
This is the error recall confidence curve confusion matrix normalized and other predictions okay we

02:21.110 --> 02:26.350
are interested in the box PR curve the precision recall curve.

02:26.390 --> 02:32.350
It's detecting all the the batteries and the precisions precision confidence curve.

02:32.670 --> 02:40.030
It's showing that all the images that contains battery are being detected with a precision confidence

02:40.070 --> 02:41.430
curve okay.

02:41.550 --> 02:44.870
So this is how to validate the model.