1 00:00:01,170 --> 00:00:06,850 In this lesson we will be using open CV to start programming with Python. 2 00:00:06,900 --> 00:00:10,080 First I want you to go to the open sea view Web site. 3 00:00:10,410 --> 00:00:19,390 Just to make sure the link is TTP s call on slash slash open CV dot org once we open the page I want 4 00:00:19,390 --> 00:00:26,200 you to go to the tutorials on the left side of the screen now want you to go to click on the main page 5 00:00:26,200 --> 00:00:36,480 tab on your left look for the link open CV Python tutorials and click it here you will see some sample 6 00:00:36,480 --> 00:00:47,100 codes using the open CV scroll down and find object detection in this lesson we will be talking about 7 00:00:47,160 --> 00:00:55,340 object detections as you see you will find many other sample codes I think that object detection sample 8 00:00:55,340 --> 00:01:02,900 is the best sample to understand most part of what we need to know when working with open CV now let's 9 00:01:02,900 --> 00:01:11,950 click on object deduction click on face detection using hard cascades here I want you to scroll down 10 00:01:12,520 --> 00:01:19,230 you can see the sample codes the object detection will enable you to get an image and recognize the 11 00:01:19,230 --> 00:01:26,680 face and the eyes here you'll see the image of the kids and how the system recognize the faces and eyes 12 00:01:28,630 --> 00:01:34,660 the process may seem complicated but if you see the sample code you can see how short of a code it is 13 00:01:35,590 --> 00:01:42,970 open CV has so much potential and that is one of the reasons it's commonly used in apps which use authentication 14 00:01:42,970 --> 00:01:45,320 technology. 15 00:01:45,460 --> 00:01:48,790 Now I want to talk about how this all works. 16 00:01:49,120 --> 00:01:55,980 Here you see the image and color in the process the image will be turned into a black and white image 17 00:01:57,180 --> 00:01:58,560 by changing the color. 18 00:01:58,590 --> 00:02:02,410 It makes the process easier to distinguish the objects in the image. 19 00:02:03,280 --> 00:02:10,020 The first step is to go through the edge features the system will search for a part which is formed 20 00:02:10,020 --> 00:02:15,090 in black and white in this gentleman's face. 21 00:02:15,120 --> 00:02:18,710 The system will see a black part in a white part right below. 22 00:02:19,440 --> 00:02:25,610 Once this color combination is detected the system will recognize that it is the eyes. 23 00:02:25,680 --> 00:02:30,780 If this feature is not detected the system will ignore this part of the image. 24 00:02:31,530 --> 00:02:39,030 If the system detects the eyes then it will go on to the next step which is the line features in this 25 00:02:39,030 --> 00:02:39,550 step. 26 00:02:39,550 --> 00:02:45,430 The system will detect a color pattern with black sides and a white in the middle. 27 00:02:45,430 --> 00:02:52,070 This will again help recognize that this part are the eyes the process will go on by narrowing down 28 00:02:52,070 --> 00:02:58,780 the image into smaller pieces until the system can recognize that the certain part are the eyes. 29 00:02:59,030 --> 00:03:02,360 And this is called the HA cascade. 30 00:03:02,360 --> 00:03:05,610 Let's scroll down to the code and see what it's doing. 31 00:03:05,630 --> 00:03:12,950 Here you can see that the example is set in the variable face cascade the color an image is converted 32 00:03:12,950 --> 00:03:21,650 into grayscale with the gray scaled image a face cascade object is set to detect where the faces if 33 00:03:21,650 --> 00:03:23,970 there are multiple faces detected. 34 00:03:24,050 --> 00:03:32,090 The process will go through a for loop twice once the face is detected it will be marked with an rectangle 35 00:03:32,660 --> 00:03:35,540 just like this image. 36 00:03:35,870 --> 00:03:42,560 Here you see eye cascade move up the code and you will find the hard cascade eye exam though has been 37 00:03:42,560 --> 00:03:43,770 read. 38 00:03:43,850 --> 00:03:49,160 Then the open see the object will detect the eyes once the eyes are detected. 39 00:03:49,310 --> 00:03:52,830 Again they would be marked with green rectangles. 40 00:03:53,540 --> 00:03:57,140 In the end the rectangles will be placed in the original image. 41 00:03:58,790 --> 00:04:04,550 Each model information to detect the face and the eyes are in these Excel files. 42 00:04:04,910 --> 00:04:08,330 So we will need to download these example files. 43 00:04:08,330 --> 00:04:18,740 You can find the XM l files and get hub go to HDTV as call on slash slash get hub dot com slash open 44 00:04:18,820 --> 00:04:31,690 CV slash open CV scroll down a bit and find data folder click on this folder and click on ha cascade. 45 00:04:31,930 --> 00:04:34,420 The first one is to detect the eyes. 46 00:04:34,660 --> 00:04:38,500 You'll see the exam L to detect the face in the middle of the list 47 00:04:41,410 --> 00:04:47,170 you can see that there are many of examples which can detect who's wearing glasses or even where a cat 48 00:04:47,200 --> 00:04:53,170 is is quite interesting so you might want to take a look of these examples. 49 00:04:53,200 --> 00:04:56,350 Now let's go on to downloading X AML files. 50 00:04:56,350 --> 00:04:59,850 First click on the hard cascade I examine. 51 00:05:00,070 --> 00:05:02,940 Click on raw to get the raw text. 52 00:05:03,310 --> 00:05:09,070 You can download by using the double get command with the link on top or you can right click on the 53 00:05:09,070 --> 00:05:13,800 site and choose save as here I'll save it to my desktop. 54 00:05:17,310 --> 00:05:25,200 Now we go back to the x amount list and choose hard cascade frontal face default x m. 55 00:05:25,340 --> 00:05:33,810 Again we click on Raw and right click to save it onto my desktop here we've completed downloading the 56 00:05:33,810 --> 00:05:35,980 example files. 57 00:05:36,030 --> 00:05:38,420 Now let's go back to the sample code. 58 00:05:38,460 --> 00:05:45,210 Now we need an image with the face the image in the sample already has the square shape so you will 59 00:05:45,210 --> 00:05:48,060 need a simple image of a face. 60 00:05:48,090 --> 00:05:54,540 Now it can be your face but if you don't want to use your face you can come to my Twitter account and 61 00:05:54,540 --> 00:05:56,310 use my face. 62 00:05:56,310 --> 00:06:06,150 My account is as a k a i j u n this is a picture of me when I used to compete in the Olympics right 63 00:06:06,150 --> 00:06:14,880 click on my image and choose said image as we can change the name to be image and save it onto my desktop. 64 00:06:14,880 --> 00:06:20,220 We've prepared the image and also the 2 x AML files to detect the face and eyes. 65 00:06:21,630 --> 00:06:24,630 Let's move onto the next lesson and start programming.