5. OpenCV Tutorial - Learn Classic Computer Vision & Face Detection (OPTIONAL)/29. Mini Project 2 - Identify Shapes (Square, Rectangle, Circle, Triangle & Stars).srt

1
00:00:00,570 --> 00:00:01,070
All right.

2
00:00:01,140 --> 00:00:04,640
So we've reached a second many projects here ship matching.

3
00:00:04,880 --> 00:00:05,860
So what we're going to do.

4
00:00:05,880 --> 00:00:11,460
We're going to take an image that looks like this here with five different images one identify each

5
00:00:11,460 --> 00:00:15,370
ship put the name of the ship inside of the ship here.

6
00:00:15,660 --> 00:00:17,910
And as we all change the color of each ship.

7
00:00:17,970 --> 00:00:19,110
So let's get started.

8
00:00:20,160 --> 00:00:20,510
OK.

9
00:00:20,540 --> 00:00:27,050
So let's get started with opening the code for a second mini projec that's lecture four point five bring

10
00:00:27,340 --> 00:00:27,970
it up here.

11
00:00:28,930 --> 00:00:29,270
OK.

12
00:00:29,310 --> 00:00:33,180
So I'll just tell you beforehand even though the code looks a bit lengthy and maybe a bit confusing

13
00:00:33,690 --> 00:00:35,140
it's actually quite simple.

14
00:00:35,460 --> 00:00:38,990
So let's run the code for this and get a idea of what's going on here.

15
00:00:39,600 --> 00:00:45,540
So D.S.O. input image here with our ship's unlabelled and everything black and white.

16
00:00:45,570 --> 00:00:45,980
There we go.

17
00:00:45,990 --> 00:00:47,160
Stuffings you one by one.

18
00:00:47,170 --> 00:00:47,910
Here's a star.

19
00:00:47,920 --> 00:00:51,540
Here's a circle triangle rectangle and a square.

20
00:00:51,540 --> 00:00:51,980
Cool.

21
00:00:51,990 --> 00:00:53,720
So let's see how it's done.

22
00:00:53,920 --> 00:00:56,870
Seriously what we do we LoDo image in here.

23
00:00:57,180 --> 00:01:03,570
As you can see when we brought up the image here there was just black and white image here.

24
00:01:03,780 --> 00:01:08,250
So it is known really the Trishul just have troubling here because it's actually become maybe about

25
00:01:08,250 --> 00:01:09,000
half that now.

26
00:01:09,270 --> 00:01:13,750
But the most important thing to do is actually to grayscale the image here.

27
00:01:13,770 --> 00:01:20,940
So loading the image here into a fine control function we use these parameters here an approximate none

28
00:01:21,360 --> 00:01:26,770
retrieve list which gives us all our contours and we have we get extractable Cantos here.

29
00:01:28,980 --> 00:01:31,590
So stepping true to us here.

30
00:01:31,740 --> 00:01:33,610
Do you remember dysfunction function approximate poly.

31
00:01:33,600 --> 00:01:40,740
DP Well what we're doing here just in case there any edges or any of the C's that are considered when

32
00:01:40,740 --> 00:01:46,270
we are finding Contos which happens quite often use approximate contours to get the approximate shape.

33
00:01:46,320 --> 00:01:48,860
So these are parameters that are fine tuned for it here.

34
00:01:49,080 --> 00:01:53,120
However feel free to justice as you see fit.

35
00:01:53,280 --> 00:01:56,260
So we have our approximate value here.

36
00:01:56,310 --> 00:01:58,120
Proximate Contos here sorry.

37
00:01:58,500 --> 00:02:04,230
So what we're doing now we're checking the length of each approximate Quanto if it's tree because that's

38
00:02:04,230 --> 00:02:10,770
all approximate that Quanto a sheep then we obviously know a tree polygon a three sided shape or a story

39
00:02:11,080 --> 00:02:12,210
shape retrieve.

40
00:02:12,350 --> 00:02:16,180
The disease is a triangle that took a while to see.

41
00:02:16,410 --> 00:02:22,740
So then we actually draw no or Cantal tell this is CND here we draw over that image and then using the

42
00:02:22,740 --> 00:02:27,870
moments just as this is just done so we can actually place our attacks in the image here and you may

43
00:02:27,870 --> 00:02:28,130
have.

44
00:02:28,160 --> 00:02:31,590
We've done this before but you may have noticed I put minus 50 here.

45
00:02:31,590 --> 00:02:39,090
And the reason they put minus 50 an x axis is because since it starts at the bottom right or bottom

46
00:02:39,090 --> 00:02:41,440
left corner the tech protects here.

47
00:02:41,460 --> 00:02:47,180
I just want to pull back the text a bit more that we had just with this text and image.

48
00:02:47,210 --> 00:02:50,340
Oh this and maybe I can do it for that for you if you want.

49
00:02:50,600 --> 00:02:56,400
But putting the minus 50 here detects was sort of like skewed across the image and it didn't look too

50
00:02:56,400 --> 00:02:57,750
nice.

51
00:02:57,750 --> 00:03:04,060
And this is a fun color we are using which is black and thickness and font size.

52
00:03:04,440 --> 00:03:05,100
So there we go.

53
00:03:05,100 --> 00:03:07,720
So that's homemade and defy if it's a triangle here.

54
00:03:08,030 --> 00:03:09,690
So pretty simple so far.

55
00:03:09,960 --> 00:03:12,070
Hopefully you follow me well.

56
00:03:12,150 --> 00:03:14,510
So what about if it's for say that shape.

57
00:03:14,720 --> 00:03:21,530
Here's what gets a bit more complicated because we have foresighted ship it can either be a square or

58
00:03:21,540 --> 00:03:22,460
rectangle.

59
00:03:22,830 --> 00:03:27,840
So again doing the Mormons and getting the bounding story getting to woman's Foose just to get labels

60
00:03:28,170 --> 00:03:28,870
done right.

61
00:03:30,230 --> 00:03:37,040
We find actually the bounding rectangle here not a bonga rectangle was done so that we can determine

62
00:03:37,190 --> 00:03:43,220
if div foresighted shape we have found here is a square or a rectangle.

63
00:03:43,220 --> 00:03:44,710
So how do we do that.

64
00:03:44,720 --> 00:03:51,280
So what do we do we basically just compare the Derwood minus to height.

65
00:03:51,290 --> 00:03:52,530
As you remember.

66
00:03:52,670 --> 00:03:56,600
When you use Boulding rectangle width and height are basically pixels.

67
00:03:56,600 --> 00:03:58,780
It extends from the x y point.

68
00:03:58,940 --> 00:04:03,470
So I imagine in most cases being x y point top left there isn't destruction.

69
00:04:03,470 --> 00:04:05,950
Height isn't destruction if it's a square.

70
00:04:06,000 --> 00:04:08,360
Those those values should be roughly similar.

71
00:04:08,630 --> 00:04:14,060
So I have here is the absolute value of W minus H is less than or equal to tree.

72
00:04:14,420 --> 00:04:16,350
Then it can be classified as a square.

73
00:04:16,700 --> 00:04:18,370
And we do it we draw it again.

74
00:04:18,410 --> 00:04:22,690
This is DiCarlo we're actually filling in the minus ones to fill the image.

75
00:04:22,780 --> 00:04:25,740
The control this is a color we're using here.

76
00:04:26,030 --> 00:04:28,210
And we're putting the text in the center again.

77
00:04:28,460 --> 00:04:30,090
So now we want to star.

78
00:04:30,470 --> 00:04:36,350
No when you actually run it to run start to it you realized is four is actually five witnesses in a

79
00:04:36,420 --> 00:04:40,960
stall and that's pretty much how we then define a style similar to the triangle.

80
00:04:41,120 --> 00:04:45,610
And again circle cycle basically we just said anything could be more.

81
00:04:45,630 --> 00:04:48,500
Anything that's more than 15 VDC is will be a signal.

82
00:04:48,500 --> 00:04:53,420
Now this could have been started that 11:12 I just have it set it for 15 minute because it gives We

83
00:04:53,420 --> 00:04:59,630
actually got more points on a star maybe maybe because I was tuning this beforehand.

84
00:05:00,800 --> 00:05:05,120
Maybe it just sort of been set 11 that doesn't matter right now because it's still identifying this

85
00:05:05,210 --> 00:05:12,500
signal because as you can imagine even with some unless we do some very non granular approximation a

86
00:05:12,530 --> 00:05:15,580
circle is going to have a lot more points and 15.

87
00:05:15,650 --> 00:05:16,340
So there we go.

88
00:05:16,430 --> 00:05:17,970
So let's run this one more time.

89
00:05:20,360 --> 00:05:23,740
And we have labeled all the images correctly here.

90
00:05:23,810 --> 00:05:29,480
So congratulations you've completed your second many projec hopefully learned a lot by using this project.