1
00:00:00,000 --> 00:00:02,190
So, congratulations on
finishing week two.

2
00:00:02,190 --> 00:00:04,530
I hope you really enjoyed
this week as much as I did,

3
00:00:04,530 --> 00:00:06,600
because image augmentation
is frankly is one of

4
00:00:06,600 --> 00:00:09,240
my favorite things to
play with in TensorFlow.

5
00:00:09,240 --> 00:00:10,845
I really enjoy the fact that

6
00:00:10,845 --> 00:00:12,510
when you have got
a small dataset,

7
00:00:12,510 --> 00:00:14,760
like this one with
the horses versus humans,

8
00:00:14,760 --> 00:00:16,770
that you're able to manipulate

9
00:00:16,770 --> 00:00:18,540
that dataset without changing

10
00:00:18,540 --> 00:00:20,550
the underlying images to open up

11
00:00:20,550 --> 00:00:21,570
whole new scenarios for

12
00:00:21,570 --> 00:00:23,850
training and to be able
to train like we saw,

13
00:00:23,850 --> 00:00:26,490
like a man lying down
or a woman facing left,

14
00:00:26,490 --> 00:00:28,290
and to be able to train
for that even if you

15
00:00:28,290 --> 00:00:30,525
don't have those images
in your training set.

16
00:00:30,525 --> 00:00:32,820
But next week, we're going
to be doing something really

17
00:00:32,820 --> 00:00:34,200
exciting and that's
called transfer

18
00:00:34,200 --> 00:00:36,520
learning. So, stay tuned.