1
00:00:01,470 --> 00:00:06,990
All right, now, let's begin our second lesson in the deep learning part of this course, where we're

2
00:00:06,990 --> 00:00:09,990
going to build a scene CNN, we just built and pay to watch.

3
00:00:10,380 --> 00:00:12,960
We're now going to use cameras with TensorFlow back end.

4
00:00:13,440 --> 00:00:17,370
So open the site Python file, which I already have done here.

5
00:00:17,880 --> 00:00:20,940
And let's take a look at this lesson now what we're going to do.

6
00:00:20,970 --> 00:00:23,490
We're going to first load the data inspected.

7
00:00:23,490 --> 00:00:25,440
Visualize it, reprocess it.

8
00:00:25,440 --> 00:00:31,560
Build our model, train our model, then plot our training logs, then save a on model, then test on

9
00:00:31,560 --> 00:00:32,820
all of us and tested it.

10
00:00:33,360 --> 00:00:36,780
A lot of these things are what we did in PyTorch in the previous lesson.

11
00:00:37,230 --> 00:00:43,560
However, not going to take a look at how we do it in Keros with TensorFlow 2.0, and you'll see that

12
00:00:43,560 --> 00:00:45,240
it's actually a lot simpler in a way.

13
00:00:45,870 --> 00:00:52,470
I hope this doesn't confuse you because getting a grasp of Python, which is actually a little bit difficult

14
00:00:52,470 --> 00:00:59,040
for beginners and TensorFlow, while easier, can sometimes confuse you and how you learn things.

15
00:00:59,430 --> 00:01:05,970
However, I think it's a very, very good idea to learn both of these libraries at the same time because

16
00:01:05,970 --> 00:01:07,710
it sort of gives you a bit better.

17
00:01:07,710 --> 00:01:16,050
And it's sort of like Ryan reinforces how you hold the concepts of how you actually learned the concepts.

18
00:01:16,560 --> 00:01:20,310
So without talking too much, let's just move ahead with this lesson.

19
00:01:20,820 --> 00:01:28,560
So firstly, in the Python lesson, we loaded our data using the Torch Vision Data Sets package and

20
00:01:28,890 --> 00:01:30,240
carries with TensorFlow.

21
00:01:30,750 --> 00:01:32,400
I'll use those terms interchangeably.

22
00:01:32,400 --> 00:01:36,600
In fact, to may decide to use Keros going forward because it's a bit easier to say.

23
00:01:37,140 --> 00:01:42,990
So with carries what we can do, we can import datasets similarly, and it can take a look and see what

24
00:01:42,990 --> 00:01:44,700
datasets are here.

25
00:01:44,850 --> 00:01:47,940
Not as much as the pie touch, but a few of the important ones here.

26
00:01:48,360 --> 00:01:52,650
These three other important ones I'll be using, as well as the F fashion list.

27
00:01:53,460 --> 00:01:54,930
So let's go back to this.

28
00:01:55,320 --> 00:01:59,280
So it's important and this all we do is just from TensorFlow.

29
00:01:59,640 --> 00:02:02,130
Don't US datasets import endless this?

30
00:02:02,610 --> 00:02:05,040
This allows us to load the amnesty to set.

31
00:02:05,370 --> 00:02:13,200
So we do amnesty of lotito and it puts the training data, the training labels, the test data and the

32
00:02:13,200 --> 00:02:17,610
test labels into these variables that will be declared here.

33
00:02:18,240 --> 00:02:21,630
So from now on, if you want to access to data, we can get it here directly.

34
00:02:22,290 --> 00:02:28,740
So importantly, it's good to check to see if we're using the GPU like we did in the previous lesson.

35
00:02:28,920 --> 00:02:32,360
So let me just firstly run this function here.

36
00:02:32,370 --> 00:02:35,160
It may take a couple of seconds to download the data.

37
00:02:35,940 --> 00:02:40,890
So while we wait for that, we'll talk about this line here so intensively.

38
00:02:41,580 --> 00:02:45,960
In order to see if you're using the GPU, you can use this quick little function here.

39
00:02:46,670 --> 00:02:53,370
Import sorry from TensorFlow to Python, the client and plot device level, and then you can use device

40
00:02:53,370 --> 00:02:55,410
level list looking devices.

41
00:02:55,890 --> 00:02:57,840
And this tells you what devices you're using.

42
00:02:58,260 --> 00:03:01,200
You can see it's a GPU is is being.

43
00:03:01,200 --> 00:03:05,550
If it's available, it will be displayed here in this part of the code here.

44
00:03:06,240 --> 00:03:13,170
So this is this was done pretty quickly with download the amnesty to set 50000 churning images at 10000

45
00:03:13,170 --> 00:03:13,950
test images.

46
00:03:14,610 --> 00:03:17,190
Let's run this code to see if you as enabled.

47
00:03:17,400 --> 00:03:22,560
If it's not, we will have to change it in the settings and it is available and to change it in the

48
00:03:22,560 --> 00:03:22,890
settings.

49
00:03:22,950 --> 00:03:28,770
It wasn't just going to edit notebook settings and change your hardware accelerated to you.

50
00:03:30,570 --> 00:03:33,300
So now let's stop there and what we'll do.

51
00:03:33,300 --> 00:03:36,180
We'll take a look at inspecting our data in the next section.

52
00:03:36,390 --> 00:03:36,810
Thank you.