WEBVTT

00:00.120 --> 00:05.920
Actually our application is detecting objects, but the labels are not correct.

00:05.920 --> 00:07.480
So this is not a vase.

00:07.720 --> 00:08.920
This is a keyboard.

00:08.920 --> 00:18.080
And this is not, uh, saying it's, um, a laptop or a computer or a PC or a screen.

00:18.080 --> 00:21.400
And this is not a book, it's the mouse.

00:21.440 --> 00:24.680
Okay, so how to fix this?

00:24.840 --> 00:27.320
Uh, that the error is in the labels.

00:27.440 --> 00:30.200
Let's open the Coco data sets labels.

00:30.360 --> 00:34.280
And by the way the Coco data set is missed.

00:34.280 --> 00:42.720
And after research for the correct Coco labels I found this, uh, this um label dot.

00:42.720 --> 00:43.640
TXT file.

00:43.960 --> 00:47.280
It contains 91, uh objects.

00:47.280 --> 00:48.760
And this is the correct one.

00:48.760 --> 00:50.440
So 91 objects.

00:50.480 --> 00:58.440
Let me copy this list and paste it here inside our Coco data set labels.

00:58.440 --> 01:00.920
And let's run our application again.

01:01.080 --> 01:02.440
And here we go.

01:02.480 --> 01:03.880
This is the keyboard.

01:03.960 --> 01:09.600
We have two keyboards and this is the monitor TV monitor.

01:09.920 --> 01:14.760
This is the laptop, this is the TV and so on.

01:14.800 --> 01:17.320
This is the mouse okay.

01:17.640 --> 01:19.080
This is another mouse.

01:19.200 --> 01:21.360
This is the book.

01:21.680 --> 01:24.840
This is a paper, but actually detecting it.

01:24.880 --> 01:25.400
Okay.

01:25.720 --> 01:28.800
So you see this is the TV monitor.

01:29.160 --> 01:35.280
This is TV, another TV monitor and this is the mouse okay.

01:35.520 --> 01:37.440
So congratulations guys.

01:37.440 --> 01:39.680
We are detecting the objects.

01:39.720 --> 01:49.720
We are succeeding in creating this amazing app that detect objects using SSD mobile.

01:49.760 --> 01:52.800
Net TensorFlow Lite model.