Original subtitles

1

Hello and welcome back to the course on computer vision then we're going to talk about some applications

2

of Ganns.

3

All right.

4

So let's have a look.

5

Gannes can be used for many different things and for instance here's a couple generating images image

6

modification super resolution assisting artists photo realistic images speech generation and face Ajahn.

7

So those are just a couple of examples of how guns can be used and we're going to look into a few of

8

them in more detail.

9

So not all of them but just a few.

10

Just to kind of like give you some ideas maybe for how you can use guides in your own life and your

11

work maybe in your own projects maybe in some applications that you might want to create.

12

All right so generating images what can guns be used for in this space.

13

Well where you saw how Gannes generate images that's kind of like the core Skase for gas or how they

14

came to be.

15

We saw that example dogs.

16

Let's have a look at a more successful example like you could see that it's really hard for a guy generative

17

adversarial network to generate an image of a like a realistic dog.

18

But let's see where they do actually.

19

Well so Franzen's these are just by looking at this can you say what this is what do you think these

20

are images of.

21

They're all images of one certain thing.

22

So you probably can already tell what they are and by the way this is just like after one iteration

23

of the one epoch.

24

And this is taken from an official paper which I will link to at the end of this tutorial but this is

25

just like a very under-trained gambit you Gary kind of see what it is.

26

If you can tell then I'm going to show you just now what it looks like what these same images look like

27

after more steps.

28

I think it was like after 12-Step so there's not steps after 12 epochs up to 12 training cycles.

29

So if we do that and let's see if you'll see if it's easier to tell what's going on.

30

So there you go.

31

So these are actually generated images and you can probably tell these are images of bedroom's.

32

And this is an example of where Gad's are doing really well some of these bedroom's looks so realistic

33

right so here you got a badge you got a pillow you've got a window.

34

Let's see what else.

35

Here you go.

36

Like a big window here somewhere and somewhere there's a bend here.

37

Here you've got a bed of some pillows and other window you can see the lighting changes the colors are

38

changing.

39

Here you go to bed and kind of like a closet over there.

40

You've got to like a bed up close over here so you can see is doing well and that's one of the cases

41

where you can do well not just bedrooms but and then just images of bedrooms but actually when you have

42

a very confined domain where something is like very you know like bedrooms on like dogs are usually

43

much more similar one to another.

44

You know they can be a of variety but there's less variety than in images of dogs because dogs are so

45

different and backgrounds and different positions are different.

46

But for instance Benham's it works quite well.

47

So we can just go back and forth a couple times so you can see just pick an image and try to observe

48

how it changes.

49

For instance look at this one you can see that it's quite pronounced here.

50

Before it was not as well defined have another one.

51

This one looks quite good.

52

It look like that.

53

I do look a little bit different as well so as you can see the more training you put in the better off

54

on our parts like these ones they don't look as defined over here.

55

But once you do more training they look more.

56

So there you go.

57

You know if you want to generate some fake images of bedrooms and gowns are your go to but also not

58

bedrooms any kind of like confined space or confined type like very tight type of image or object that

59

where the variety isn't that great and gowns can do a pretty good job image modification.

60

OK so what does that mean.

61

Well let's have a look at this example.

62

So here we've got this from the same paper and here we got a smiling woman on the left.

63

This is these are images generated by Gahn all of them.

64

Then you go to a neutral woman on in the middle and then you go to a neutral man.

65

So if you take the smiling woman.

66

So basically what they're showing here is that through Gannes through the way that these images are

67

generated they're actually assigned a vector.

68

If you look at the neural network that represents them and represent that in a vector you can then do

69

arithmetic arithmetics with those vectors so you can take a picture of a smiling woman and subtract

70

the vector for a neutral woman and you'll get.

71

And then you add a veteran for unusual man.

72

Well you you'll get the smiling man.

73

So basically by subtracting this from this you'll get a vector for the smile you added on top of the

74

natural man and you get them one that is smiling.

75

Very very interesting.

76

A little bit creepy as well.

77

But you know that's they stayed there.

78

Now here's another example.

79

Manner of glasses subtract magnifying glasses completely different man as you can see.

80

And then you add a woman with fogged glasses and you get a woman with glasses.

81

Very interesting and so similar results have been already seen in the space of context where for instance

82

a very famous example is if you take her use neural networks to her present words or like sentences

83

and then you take or like the word sorry if you just take send.

84

If you take words and represent them vectors and then for instance you take king and you subtract man

85

and you add the word woman then you get something close to a queen and that can be that has been achieved

86

before.

87

But it's not as like as not as big as a breakthrough as the same with images because images actually

88

represent not just something that we have like an alphabet or different words actually represent something

89

I can see in real life and you can see this movement of glass all the way here is pretty distinct.

90

It's pretty you can really see that it's warm and glossy even though these people never existed.

91

This is just all gang generated images.

92

And for example here they're showing that if you try to do the same with just arithmetic operations

93

for pixels then you will never actually have the result as good because as you can see you just get

94

like overlay on different images and that's nowhere near to as good as what we had over here.

95

All right.

96

So that's what does this.

97

What does this application image modification.

98

So this next one is super resolution.

99

Here's an example from the end Goodfellows presentation where they had an original image which looked

100

like that.

101

Then they reduced the resolution so made it like well like a more grainy and less detail.

102

And then they applied different algorithms to increase their resolution back.

103

And so that and by Kubik there is as far as s.

104

NET and this superposition by again.

105

And so you can see that that can do a really good job even though this was a bit smoother and a really

106

good job and you can see that it's quite Kristo.

107

Chris backflushing they could really not really call if they had the image of the the the low resolution

108

that they created in between here so the image that they took from this what they created and then the

109

results would be called to compare but I couldn't find that image unfortunately.

110

So all we have to compare is to the original.

111

And we can see kind of like how this could perform over well overall well but if you look in detail

112

like for instance here on the hat there is a pattern and it's not here.

113

There are some like horses or elephants there that are not here.

114

There's certain detail here that has been destroyed but nevertheless this is a bear of them.

115

You could see Prius's So now imagine if you didn't have the original if you just have a low resolution

116

image and you want to recreate the higher resolution version of that image then then you could use a

117

gown for that as well.

118

OK.

119

Photo realistic images.

120

This is a very interesting one.

121

So it was this company owner company was this application online pics depicts which allows you to like

122

you draw something with just on line and then you apply and generative adversarial literally it just

123

presses button basically and you get a photo realistic image so you get.

124

So again he uses this as an input and learns what a human should look like.

125

He knows kind of from it's training where a human should look like and overlay styles on top of that

126

overlay.

127

Turn this image into something that looks like a real photo.

128

And this service was so popular they got over like 2 million views 2 million visitors on the Web site.

129

The people who put it up there they just didn't have enough funds to maintain the service so they had

130

to close that unfortunately.

131

But there's a really cool YouTube video of a an artist playing around with this and we'll link to it

132

at the end.

133

In additional reading or viewing in this case and you'll be able to check it out there's some very interesting

134

stuff.

135

But this is really crazy like from just a pencil drawing a line drawing you get a realistic looking

136

photo like that face aging.

137

So a popular one so I think these are all Hollywood actors.

138

I'm not actually sure they look like Hollywood action so.

139

And I don't really know the background of this how they did it did they lose.

140

Maybe they took these photos of here at the start and then age them.

141

But basically the word was applied here is a generator about a central network to generate a sequence

142

of photos from the photo.

143

So basically you can use you sort of like a young photo and see how a person will age over time by applying

144

Genter of adversarial networks.

145

Of course they have to go through certain training to understand you know the concepts of aging too

146

for those concepts to be programmed into the nor embedded into the neural network.

147

But this is another application of general advertising in general.

148

There are several networks as you can see quite a lot of different applications.

149

So the couple of resources that I mentioned this one is a paper called as it provides representation

150

learning with deep convolutional generation generative adversarial networks so DC Gannes This is like

151

a step above Gannes or an enhancement on top of Gannes whether networks or even deeper.

152

And you can find it on our archive there and it's actually going to a lot of those examples are we talked

153

about.

154

They are given in this paper if you want to read about it and see some additional images and the YouTube

155

that are referenced is called artist versus picks topics.

156

Is this humor or horror.

157

Check it out.

158

This just this art is drawing these images.

159

If you just Google artist versus Bicks depicts you'll find this artist drawing these images and then

160

applying the Gahn to see how it will overlay human skin and hair and tones over and.

161

Very interesting all of the time it's very creepy.

162

All right so that's a precaution of Ganze.

163

Hope you enjoyed this Tauriel and courtsey next time.

164

Until then enjoy computer vision.