1 00:00:08,190 --> 00:00:10,640 Hey guys this is Caleb with slopes dot com. 2 00:00:10,640 --> 00:00:15,640 And in this video we're going to dive right into adding our core a male model. 3 00:00:15,660 --> 00:00:20,850 We're going to pass an image into it and actually use machine learning to make a prediction of what 4 00:00:20,850 --> 00:00:22,760 it thinks that it's seeing in the image. 5 00:00:22,770 --> 00:00:24,080 Very very cool. 6 00:00:24,120 --> 00:00:30,440 In the last video we left off with this we have our camera view we have our nice labels and our UI views 7 00:00:30,450 --> 00:00:36,380 and when we tap the screen it takes a photo of what it sees and it saves it to the image of you. 8 00:00:36,450 --> 00:00:41,360 We're going to use that in just a moment to pass into our core a male model. 9 00:00:41,380 --> 00:00:47,400 OK so to begin we're actually going to pull open Safari and we're going to go ahead and head to the 10 00:00:47,400 --> 00:00:49,560 core M-L site from Apple. 11 00:00:49,560 --> 00:00:54,860 If you don't remember it's app developers Apple dot com slash machine learning. 12 00:00:54,930 --> 00:01:01,950 Now if you scroll way down to the bottom there is a section called working with models and basically 13 00:01:01,950 --> 00:01:08,580 Apple have provided quite a few different machine learning models that work with core m-L. 14 00:01:08,730 --> 00:01:11,620 The one we're going to use is called squeeze net. 15 00:01:11,820 --> 00:01:18,630 And what it does is it detects the dominant objects present in an image from a set of 1000 categories 16 00:01:19,020 --> 00:01:27,240 like trees animals food vehicles people and more but with an overall footprint of 4.7 megabytes it's 17 00:01:27,240 --> 00:01:28,590 really really small. 18 00:01:28,650 --> 00:01:33,270 And if you actually look at some of these other core and L models you'll see they're pretty big 100 19 00:01:33,270 --> 00:01:35,270 to megabytes 94 megabytes. 20 00:01:35,270 --> 00:01:37,770 Five hundred fifty three megabytes. 21 00:01:37,770 --> 00:01:42,540 So this one is just way more efficient and way smaller. 22 00:01:42,540 --> 00:01:46,260 I mean that's just good for our app it will make the apps size much smaller in the end. 23 00:01:46,260 --> 00:01:51,750 So to get that model go ahead and just click download core Armel model and it'll save to your downloads 24 00:01:51,750 --> 00:01:53,280 which you can see right here. 25 00:01:58,650 --> 00:02:01,930 Which you can see right here and it is called squeeze net. 26 00:02:02,050 --> 00:02:06,650 M-L model that of course stands for machine learning model. 27 00:02:06,660 --> 00:02:11,820 Now what we need to do is open our X called Project and add this file into the project so go ahead and 28 00:02:11,820 --> 00:02:18,030 pull open your X code project and I'm actually going to just drag this from safari and drop it right 29 00:02:18,030 --> 00:02:19,830 into my project. 30 00:02:19,860 --> 00:02:26,910 Now of course copy items if needed create groups and make sure that the target vision app has been selected. 31 00:02:27,050 --> 00:02:33,480 OK click Finish and we're going to go check out what is in this M-L model so click on that and you'll 32 00:02:33,480 --> 00:02:36,410 notice here it says that it is a machine learning model. 33 00:02:36,490 --> 00:02:39,280 It's of type and neural network classifier. 34 00:02:39,300 --> 00:02:46,710 So now you understand how the structure is is working and it gives you a nice description and also references 35 00:02:46,710 --> 00:02:47,700 the inputs. 36 00:02:47,700 --> 00:02:49,160 OK this is important to know. 37 00:02:49,410 --> 00:02:58,210 So we input an image and you see right here that it is BGR 6:58 and you can't see it but that's 6:58. 38 00:02:58,320 --> 00:03:05,720 That means that the input image that it can classify is 227 pixels by 227 pixels. 39 00:03:05,940 --> 00:03:08,530 The outputs are what we get at the end. 40 00:03:08,530 --> 00:03:13,710 OK so we get class label probs which is the probability of each category. 41 00:03:13,890 --> 00:03:18,020 And we also get the class label which is the most likely image category. 42 00:03:18,120 --> 00:03:23,130 So it'll tell us what it thinks it is and the probability that it thinks it's correct. 43 00:03:23,160 --> 00:03:28,380 If you notice here we have squeezed net and it says that it is a swift generated source. 44 00:03:28,530 --> 00:03:35,370 That means that squeeze net is accessible to us as as a model like we can use it just like we can use 45 00:03:35,400 --> 00:03:36,680 any other class. 46 00:03:36,810 --> 00:03:39,550 And let's click this little arrow here so I can show you that. 47 00:03:39,570 --> 00:03:40,200 Check it out. 48 00:03:40,320 --> 00:03:47,470 So it is a Class A swift class it's ready for us to use with squeeze net input right here squeeze net 49 00:03:47,520 --> 00:03:49,230 output can scroll down. 50 00:03:49,230 --> 00:03:55,780 We can actually see the whole squeeze net class right here it's very cool and all available for us. 51 00:03:55,800 --> 00:03:59,880 That way you can take a look at this more and you can kind of dive in deeper if you want to but for 52 00:03:59,880 --> 00:04:05,070 now we're just going to go ahead and we're going to add this into the resources folder and go to camera 53 00:04:05,060 --> 00:04:05,830 AVC. 54 00:04:05,850 --> 00:04:06,840 Let's use it. 55 00:04:06,840 --> 00:04:08,820 Let's see what it can do. 56 00:04:08,840 --> 00:04:12,720 So to begin we're going to go ahead and we're going to actually import two frameworks. 57 00:04:12,720 --> 00:04:17,440 First we're going to import core M-L and last we're going to import vision. 58 00:04:17,560 --> 00:04:24,870 OK these are both part of the core M-L package Corum El handles just kind of your basic overarching 59 00:04:24,930 --> 00:04:32,730 all of the machine learning stuff and vision specifically handles things like face recognition and object 60 00:04:32,730 --> 00:04:33,420 recognition. 61 00:04:33,450 --> 00:04:37,730 We already talked about that and one of the first videos but just just a quick reminder. 62 00:04:37,740 --> 00:04:43,710 All right so now we get to actually use this machine learning model we've dragged in Squeeze net and 63 00:04:43,710 --> 00:04:49,200 to do that go ahead and scroll down and where we created the photo data we're actually going to use 64 00:04:49,200 --> 00:04:54,690 that same photo data and pass it into our core email model squeeze net. 65 00:04:54,780 --> 00:05:01,280 So to do that go ahead and go beneath the photo data and we're going to use a do try catch block k. 66 00:05:01,500 --> 00:05:08,340 We need to be able to catch any errors that are thrown and using vision can throw errors and so we're 67 00:05:08,340 --> 00:05:12,540 going to set it up so that it can handle that and not cause a crash in our application. 68 00:05:12,540 --> 00:05:13,960 So go ahead and type do. 69 00:05:14,460 --> 00:05:17,660 And we're going to just set up the catch block as well while we're at it. 70 00:05:18,060 --> 00:05:18,630 Perfect. 71 00:05:18,630 --> 00:05:22,310 And this is where we're going to handle errors in just a moment. 72 00:05:22,440 --> 00:05:25,590 But inside the do block we're going to type let. 73 00:05:25,590 --> 00:05:31,180 Model equals try v and core Corum L model k. 74 00:05:31,200 --> 00:05:35,890 Now we're going to do a parentheses and we're going to pass in an L model. 75 00:05:35,910 --> 00:05:42,570 Now if you remember squeeze net is an M-L model so what we can do is we can just pass in Squeeze net 76 00:05:42,990 --> 00:05:48,180 we can instantiate it and if we put a period we can pull out the male model. 77 00:05:48,180 --> 00:05:55,560 If you go to look at this this class you can see that there is a variable here that is for an N O model 78 00:05:55,570 --> 00:06:00,030 so we can pull that out and we can use that as a vision core I.M. model. 79 00:06:00,030 --> 00:06:00,840 Really really cool. 80 00:06:00,840 --> 00:06:06,840 So basically we're using squeezed now but we're passing it through vision because vision has a lot of 81 00:06:06,840 --> 00:06:09,400 really cool stuff for identifying things and images. 82 00:06:09,650 --> 00:06:10,890 Okay. 83 00:06:10,980 --> 00:06:16,200 And of course we're using tri because Wii and Corum male model can throw errors and you'll notice in 84 00:06:16,200 --> 00:06:20,000 a while little say that we're having some errors because we're not properly handling them in our catch 85 00:06:20,000 --> 00:06:20,300 block. 86 00:06:20,310 --> 00:06:21,420 But we will. 87 00:06:21,450 --> 00:06:23,370 So we now have a model. 88 00:06:23,370 --> 00:06:24,290 This is good. 89 00:06:24,300 --> 00:06:30,480 The models like our brain and now we need to make a request which is kind of like a thought case where 90 00:06:30,810 --> 00:06:32,820 we're allowing our brain to have a thought. 91 00:06:32,850 --> 00:06:34,990 And then it's going to do something with that thought later. 92 00:06:34,980 --> 00:06:39,640 So after we've created our model go ahead and type let request. 93 00:06:39,840 --> 00:06:43,460 And this is where we're going to use a Vienne core M-L request. 94 00:06:43,460 --> 00:06:45,230 Gates right there ready for you. 95 00:06:45,330 --> 00:06:48,150 Now if you put up parentheses you can see that there are two options. 96 00:06:48,150 --> 00:06:52,740 There's one with a model and one with the completion handler and model. 97 00:06:52,740 --> 00:06:59,390 We're going to use that one because our completion handler will be where we will handle what happens 98 00:06:59,390 --> 00:07:02,110 when we get a response back from this request. 99 00:07:02,120 --> 00:07:02,580 OK. 100 00:07:02,690 --> 00:07:06,170 Now our model of course is going to be our model. 101 00:07:06,170 --> 00:07:12,090 And now this completion handler This is where we're going to call a function we will right in a moment. 102 00:07:12,260 --> 00:07:18,240 So for now I'm going to go ahead and write that function but I'm not going to use it yet. 103 00:07:18,410 --> 00:07:25,400 So up here in our camera AVC class we need to write that function and we're going to put it beneath 104 00:07:25,400 --> 00:07:26,910 did tap camera view. 105 00:07:27,050 --> 00:07:32,700 So I'm going to go ahead and type phunk results method. 106 00:07:32,800 --> 00:07:33,150 All right. 107 00:07:33,160 --> 00:07:33,410 Oops. 108 00:07:33,410 --> 00:07:37,220 And we're going to need to pass in a request of V.N. request. 109 00:07:37,220 --> 00:07:43,760 That's a vision request and we need to also give it the capability of using the errors that could come 110 00:07:43,760 --> 00:07:49,380 from our from our request so that's going to be of type error. 111 00:07:49,880 --> 00:07:51,480 And there isn't always an error. 112 00:07:51,530 --> 00:07:53,660 That's why we're creating it as an optional. 113 00:07:53,660 --> 00:07:58,270 All right so go ahead and give it some curly braces and we will come back to handle 114 00:08:00,540 --> 00:08:02,880 changing the label text. 115 00:08:02,880 --> 00:08:07,460 OK we'll come back to that in a moment but right now we just need to be able to use result's method. 116 00:08:07,620 --> 00:08:08,280 OK. 117 00:08:08,520 --> 00:08:11,840 So our completion handler is going to be result's method. 118 00:08:12,270 --> 00:08:20,220 And what we're going to do is we're going to basically just call the name of our function here because 119 00:08:20,670 --> 00:08:26,330 when we create a request it automatically has a V.N. request and it automatically has an error. 120 00:08:26,490 --> 00:08:31,320 So that can just be passed in by default to this function so we don't actually need to put in any of 121 00:08:31,320 --> 00:08:32,040 the parameters. 122 00:08:32,040 --> 00:08:34,210 That's kind of the cool thing about using this. 123 00:08:34,280 --> 00:08:36,600 So we have a model that's our brain. 124 00:08:36,600 --> 00:08:42,210 We have a request that's kind of like the our brain is making a thought and now the handler is going 125 00:08:42,210 --> 00:08:46,760 to basically take that thought and it's going to try to turn it into something that we can use. 126 00:08:46,770 --> 00:08:51,770 So let's go ahead and let's create a handler by typing let hndler. 127 00:08:52,260 --> 00:08:57,370 And that's going to be an image request handler. 128 00:08:57,390 --> 00:08:58,400 OK. 129 00:08:58,440 --> 00:09:00,320 Now of course there are lots of options here. 130 00:09:00,330 --> 00:09:08,180 As you can see CIA image C.G. image C-v pixel buffer data and you are l what we're going to do is we're 131 00:09:08,190 --> 00:09:13,900 actually going to go ahead and just click on the one with data and options is necessary. 132 00:09:13,920 --> 00:09:15,350 We don't need to use the options. 133 00:09:15,390 --> 00:09:16,280 So we'll get rid of it. 134 00:09:16,500 --> 00:09:18,590 But we are going to pass in our data. 135 00:09:18,750 --> 00:09:19,960 What data you ask. 136 00:09:20,040 --> 00:09:21,260 Our photo data. 137 00:09:21,300 --> 00:09:21,850 OK. 138 00:09:22,050 --> 00:09:26,640 That's the data we're going to give it so that it can analyze and determine what's in the image so pass 139 00:09:26,650 --> 00:09:31,510 in photo data and go ahead and unwrap it so that we can use it. 140 00:09:31,530 --> 00:09:37,350 Now what we need to do is we basically need to use our handler to perform our request our handler is 141 00:09:37,350 --> 00:09:40,590 kind of like a part of our brain that handles our ability to think. 142 00:09:40,590 --> 00:09:45,120 So our handler is going to take our thought here and it's going to try to process it. 143 00:09:45,240 --> 00:09:45,740 OK. 144 00:09:45,990 --> 00:09:53,370 So let's go ahead and we're going to actually need to use try as well because the handler can also throw 145 00:09:53,370 --> 00:09:59,610 errors and so let's go ahead and type try handler perform and you see right here. 146 00:09:59,620 --> 00:10:02,460 It can take an array of V.N. requests. 147 00:10:02,880 --> 00:10:07,690 And so we're going to pass it an array with our request. 148 00:10:07,720 --> 00:10:08,030 OK. 149 00:10:08,040 --> 00:10:12,170 So if we try to build and run here you'll notice it will work. 150 00:10:12,180 --> 00:10:14,720 But the issue is that we're not handling the errors. 151 00:10:14,730 --> 00:10:16,670 If there's a problem we should know. 152 00:10:16,890 --> 00:10:18,720 So to do that we're just going to use Debug. 153 00:10:18,720 --> 00:10:20,350 Print like we have been doing. 154 00:10:20,700 --> 00:10:23,040 And we're going to just call the error. 155 00:10:23,100 --> 00:10:24,080 There's an error. 156 00:10:24,120 --> 00:10:25,270 It'll handle it here. 157 00:10:25,540 --> 00:10:25,760 OK. 158 00:10:25,800 --> 00:10:27,450 That's really easy. 159 00:10:27,450 --> 00:10:35,280 So now what we're doing is we're basically saying that we have a model OK we are going this is like 160 00:10:35,280 --> 00:10:37,650 our brain can process the image. 161 00:10:37,650 --> 00:10:41,540 We have a request that is using our brain to make a thought. 162 00:10:41,670 --> 00:10:47,710 We have a handler that's going to basically take the data and compare it against our little thought. 163 00:10:47,790 --> 00:10:53,310 And then it's going to perform that request work compares them and produces some data for us that is 164 00:10:53,370 --> 00:10:56,980 actually all we need to do to use machine learning. 165 00:10:57,000 --> 00:11:03,610 For now we're going to actually pull out the information from our request in this function results method. 166 00:11:03,810 --> 00:11:08,250 And that's how we're going to actually display what our M-L model is thinking. 167 00:11:08,430 --> 00:11:15,390 So let's do that now go to results method and what we're going to do is we're going to create a constant 168 00:11:15,630 --> 00:11:18,060 to hold the results that come in from our request. 169 00:11:18,060 --> 00:11:21,910 Remember it just gets passed in straight here through our completion handler. 170 00:11:21,930 --> 00:11:28,980 So get rid of this commented out words it's not code and we're going to use guard Latt to create an 171 00:11:28,980 --> 00:11:32,520 instance of our results so go ahead and type guard. 172 00:11:32,520 --> 00:11:36,610 Let results and that is going to be equal to request. 173 00:11:36,630 --> 00:11:43,670 And if you type the period after that you can see that there is results inside of a request. 174 00:11:43,710 --> 00:11:49,950 So very cool we can get that but it needs to be we need to put it into an array because we're going 175 00:11:49,950 --> 00:11:55,160 to cycle through it and we're going to check to make sure that we get the most confident result. 176 00:11:55,160 --> 00:11:56,440 We don't want. 177 00:11:56,750 --> 00:11:58,760 It's like one tenth of a percent sure. 178 00:11:58,770 --> 00:12:02,160 We don't really care about that we want something that is almost positive. 179 00:12:02,160 --> 00:12:10,950 So go ahead and we're going to cast this as an array of V.N. classification observation and you know 180 00:12:10,950 --> 00:12:17,010 what you guys can dive deeper into these descriptions for what these are but basically it is seen classification 181 00:12:17,010 --> 00:12:20,580 information produced by an image analysis request. 182 00:12:20,580 --> 00:12:21,360 All right. 183 00:12:21,630 --> 00:12:28,130 So it's basically using it to determine what is in the image that it was sent in the request. 184 00:12:28,350 --> 00:12:32,260 So with a guard left at the end we need to type else. 185 00:12:32,520 --> 00:12:34,770 And then inside of here it's pretty common just to return. 186 00:12:34,770 --> 00:12:39,900 So if it doesn't work if we if we don't get any results we're just going to return so that we don't 187 00:12:39,900 --> 00:12:46,260 have a crash if we do then we have a nice constant here called results that has our results. 188 00:12:46,320 --> 00:12:48,630 So that looks great. 189 00:12:48,630 --> 00:12:56,930 And beneath result's we're going to go ahead and type for classification in results. 190 00:12:56,940 --> 00:13:01,990 And the reason for that is because we are using V.N. class an observation and that's what we're doing 191 00:13:01,990 --> 00:13:08,050 is we're we're classifying objects in an image based on criteria that the M-L model understands. 192 00:13:08,050 --> 00:13:12,000 So for each classification let's think what we want to do here. 193 00:13:12,520 --> 00:13:18,700 What we're going to do is basically we're going to set it up so that if it is more than 50 percent confident 194 00:13:19,540 --> 00:13:22,180 then it's going to say what it thinks it is. 195 00:13:22,210 --> 00:13:26,680 If it's less than 50 percent confident it's going to say I'm not really sure what this is. 196 00:13:26,680 --> 00:13:27,840 Please try again. 197 00:13:28,090 --> 00:13:29,260 Simple as that. 198 00:13:29,290 --> 00:13:35,300 So to do that we're going to go ahead and type if classification confidence. 199 00:13:35,470 --> 00:13:37,010 Now you notice that. 200 00:13:37,210 --> 00:13:44,900 Well I might have to actually just a second if classification confidence is less than 0.5 K 50 percent. 201 00:13:45,010 --> 00:13:50,140 And if I select this you can see that V.N. confidence is of type float. 202 00:13:50,220 --> 00:13:50,610 OK. 203 00:13:50,710 --> 00:13:52,270 It's a float it's a number that comes in. 204 00:13:52,270 --> 00:13:58,510 So we have 50 percent here 0.5 50 hundredths or five tenths. 205 00:13:58,780 --> 00:14:05,860 If it is less than 50 percent what we're going to do is we're going to set the text label here in our 206 00:14:06,700 --> 00:14:10,540 main upper bar we're going to set that to say I'm not sure what this is. 207 00:14:10,540 --> 00:14:11,700 Please try again. 208 00:14:12,010 --> 00:14:21,560 So go ahead and let's type self dot identification label text and that's going to be equal to. 209 00:14:21,790 --> 00:14:23,790 I'm not sure what this is. 210 00:14:23,800 --> 00:14:26,650 Please try again. 211 00:14:26,650 --> 00:14:26,960 All right. 212 00:14:26,980 --> 00:14:27,610 And you know what. 213 00:14:27,610 --> 00:14:32,570 If we don't have if we're not really confident we shouldn't even show this confidence label so Gordon 214 00:14:32,590 --> 00:14:37,990 type self-confidence label that text and we're just going to say that to be an empty string because 215 00:14:37,990 --> 00:14:39,060 there's no need to show it. 216 00:14:39,070 --> 00:14:45,700 If there is not much confidence that it actually works OK and if that happens as well there's not really 217 00:14:45,760 --> 00:14:48,930 a need for it to continue so we're just going to call break. 218 00:14:48,940 --> 00:14:51,350 So that leaves the for loop. 219 00:14:51,440 --> 00:14:52,270 OK. 220 00:14:52,810 --> 00:14:58,230 What we're going to do next is what happens if we do get the proper results. 221 00:14:58,240 --> 00:15:02,800 Let's say that we get something it thinks it's 98 percent sure that it just saw an umbrella. 222 00:15:02,880 --> 00:15:04,930 K so else right. 223 00:15:05,040 --> 00:15:10,050 That means it's greater than 50 percent if the confidence is more than 50 percent. 224 00:15:10,060 --> 00:15:17,110 Go ahead and type self identification label dot text and that's going to be equal to the classification 225 00:15:18,010 --> 00:15:19,380 identifier. 226 00:15:19,410 --> 00:15:21,560 The identifier is what it thinks it saw. 227 00:15:21,570 --> 00:15:27,760 So in the example I showed you at the beginning of this course when you tap it takes a photo and it 228 00:15:27,760 --> 00:15:29,430 shows maybe umbrella. 229 00:15:29,830 --> 00:15:33,520 And we will program it later to speak out loud to say what it thinks it is. 230 00:15:33,520 --> 00:15:41,050 But this identifier is a string and it returns a name of an object or an item and that's all we need 231 00:15:41,050 --> 00:15:47,400 to do next we need to go ahead and set our confidence label to show how confident it thinks it is. 232 00:15:47,410 --> 00:15:54,490 So go ahead and type self confidence label that text and that's going to be equal to a string of course 233 00:15:54,490 --> 00:16:01,270 and we need to type confidence just like we have here with the capital letters confidence and now we're 234 00:16:01,270 --> 00:16:06,130 going to use string interpolation to pass in a value that is not a string but we're going to cast it 235 00:16:06,130 --> 00:16:07,820 as a string so that it can be used. 236 00:16:07,840 --> 00:16:10,610 So we're going to use the classification. 237 00:16:10,780 --> 00:16:12,270 Confidence. 238 00:16:12,330 --> 00:16:12,610 OK. 239 00:16:12,610 --> 00:16:13,900 Pretty cool. 240 00:16:13,900 --> 00:16:20,260 And we're going to actually multiply that by 100 because right now if the value came in at 0.5 that 241 00:16:20,260 --> 00:16:22,620 might look like 0.5 percent sure. 242 00:16:22,640 --> 00:16:23,380 We don't want that. 243 00:16:23,380 --> 00:16:29,440 So if we were to multiply this by 100 it would move the decimal place over and it would say 50 percent. 244 00:16:29,440 --> 00:16:29,700 OK. 245 00:16:29,740 --> 00:16:33,720 So that's what we want because 0.5 is 50 percent. 246 00:16:33,820 --> 00:16:38,270 Now the issue is that it would be fifty point zero percent. 247 00:16:38,350 --> 00:16:43,030 And I don't want to have to deal with the additional numbers on the outside. 248 00:16:43,030 --> 00:16:48,520 So I'm actually going to cast this as an integer and then put another parentheses on the outside so 249 00:16:48,520 --> 00:16:50,450 that this number is a whole number. 250 00:16:50,590 --> 00:16:51,550 51 percent. 251 00:16:51,550 --> 00:16:53,220 37 percent. 252 00:16:53,290 --> 00:16:57,400 If you want you can leave it at thirty seven point nine eight seven six five percent sure. 253 00:16:57,400 --> 00:17:02,210 I just got kind of annoyed reading like eight numbers after the decimal place so I cast it as an ant. 254 00:17:02,230 --> 00:17:04,720 That's just my personal preference. 255 00:17:04,840 --> 00:17:07,470 And let's see confidence we have the classification. 256 00:17:07,480 --> 00:17:10,870 And at the very end we need to add a percentage mark. 257 00:17:10,890 --> 00:17:12,150 All right. 258 00:17:12,310 --> 00:17:14,360 That looks pretty good. 259 00:17:15,070 --> 00:17:17,920 You know what do you guys want to go try it let's go see if it worked. 260 00:17:18,160 --> 00:17:23,230 OK we're going to go ahead and build and run it and it will show up here as soon as it builds and runs 261 00:17:23,230 --> 00:17:25,030 it will close. 262 00:17:25,210 --> 00:17:25,690 Maybe. 263 00:17:25,840 --> 00:17:26,950 Yeah there it is. 264 00:17:27,310 --> 00:17:27,960 All right. 265 00:17:28,150 --> 00:17:34,910 Building and installing to my phone it'll pop open. 266 00:17:34,920 --> 00:17:36,370 There it is. 267 00:17:36,390 --> 00:17:36,720 All right. 268 00:17:36,720 --> 00:17:38,730 Should we try it if it worked. 269 00:17:38,730 --> 00:17:40,160 We should tap the screen. 270 00:17:40,290 --> 00:17:45,750 It should analyze the photo and it should predict what it thinks it is and return a level of confidence. 271 00:17:45,750 --> 00:17:46,200 Let's try. 272 00:17:46,200 --> 00:17:46,410 Ready. 273 00:17:46,410 --> 00:17:48,650 Three two one. 274 00:17:50,590 --> 00:17:51,760 I'm not sure what this is. 275 00:17:51,760 --> 00:17:54,250 Please try again. 276 00:17:54,310 --> 00:17:55,450 I'm not sure what this is. 277 00:17:55,450 --> 00:17:56,700 Please try again. 278 00:17:58,700 --> 00:17:59,260 Ha. 279 00:17:59,480 --> 00:18:01,880 Interesting. 280 00:18:01,900 --> 00:18:04,030 So it does not appear to know what this is. 281 00:18:04,030 --> 00:18:10,940 Let's try some other other things maybe it might know what my wallet is my desk Loeb's wallet OK. 282 00:18:11,130 --> 00:18:12,650 Not sure what this is interesting. 283 00:18:12,660 --> 00:18:15,530 So what is the problem here. 284 00:18:16,870 --> 00:18:24,340 Let's see we're setting the label you know what let's go ahead let's actually print out the classification 285 00:18:24,340 --> 00:18:24,910 results 286 00:18:29,090 --> 00:18:30,930 let's print out the identifier. 287 00:18:30,960 --> 00:18:35,670 This way we can see everything that it's thinking through us that we know if it's 288 00:18:38,710 --> 00:18:38,960 yeah. 289 00:18:39,040 --> 00:18:40,450 So we can get an idea of what it's thinking. 290 00:18:40,450 --> 00:18:45,290 So when I tap the screen oh it says remote control and modem. 291 00:18:45,310 --> 00:18:50,320 I wonder why oh ok so I think I know what the problem is. 292 00:18:50,320 --> 00:18:52,530 So basically what we're doing is we're going through. 293 00:18:52,690 --> 00:18:58,000 If the confidence is less than 50 percent basically we go through we say I don't know what this is. 294 00:18:58,000 --> 00:19:00,490 And we break meaning we leave our for loop. 295 00:19:00,730 --> 00:19:06,880 But then if it is more than 50 percent confident it goes here and it identifies this but then it loops 296 00:19:06,880 --> 00:19:09,620 through again and it hits something that it's not confident on. 297 00:19:09,640 --> 00:19:14,740 And it shows I'm not sure what this is so when we hit something that it is confident on more than 50 298 00:19:14,740 --> 00:19:16,340 percent we should break. 299 00:19:16,390 --> 00:19:17,350 All right. 300 00:19:17,770 --> 00:19:20,360 OK so we should go ahead and do that. 301 00:19:20,740 --> 00:19:27,190 Let's break on both so that if it's not confident it breaks if it is not if it is totally confident 302 00:19:27,190 --> 00:19:30,390 on the first one it will just go right to here and that's fine. 303 00:19:30,390 --> 00:19:32,390 So let's let's build and run. 304 00:19:32,470 --> 00:19:37,180 Basically we were just cutting out our most confident result because we were going through some of the 305 00:19:37,210 --> 00:19:38,710 non-confidence ones first. 306 00:19:38,710 --> 00:19:41,780 So now that it's built let's take a look. 307 00:19:41,830 --> 00:19:48,310 When I tap on the screen to look at that remote control it's ninety nine percent confidence. 308 00:19:48,320 --> 00:19:53,670 Let's scan something that maybe it doesn't know maybe it doesn't know a Rubik's cube. 309 00:19:53,670 --> 00:19:54,390 Let's try it. 310 00:19:55,470 --> 00:19:55,750 OK. 311 00:19:55,770 --> 00:19:56,730 I'm not sure what this is. 312 00:19:56,730 --> 00:19:57,300 Please try again. 313 00:19:57,300 --> 00:20:00,540 That's fine because I doubt it was trained to know what a Rubik's cube is. 314 00:20:00,600 --> 00:20:01,700 So that's good to know. 315 00:20:01,890 --> 00:20:06,410 When I scanned the remote again remote control 97 percent confident. 316 00:20:06,420 --> 00:20:07,190 Very cool. 317 00:20:07,200 --> 00:20:08,970 So it looks like it's working. 318 00:20:08,970 --> 00:20:10,250 This is awesome. 319 00:20:10,290 --> 00:20:13,140 So we have just successfully integrated machine learning. 320 00:20:13,140 --> 00:20:19,650 I kid you not it's that easy to pass in images make predictions based on a trained model we just gave 321 00:20:19,650 --> 00:20:26,430 our app the ability to think how incredible is that amazing work in the next video we're actually going 322 00:20:26,430 --> 00:20:31,460 to set it up so that we can toggle the flash to turn on and off. 323 00:20:31,650 --> 00:20:36,870 And then we're going to actually at the very end we're going to set up a an avi speech synthesizer to 324 00:20:36,870 --> 00:20:42,720 speak the results to us thinking about maybe users who might be vision impaired who might not be able 325 00:20:42,720 --> 00:20:48,310 to see very well but want to classify objects they can hear what the app is seeing. 326 00:20:48,390 --> 00:20:49,350 Very very cool. 327 00:20:49,350 --> 00:20:52,100 So we're going to go ahead and move on to the next video. 328 00:20:52,110 --> 00:20:57,600 Amazing work with this one guys we are on our way to have a full fledged machine learning app that can 329 00:20:57,600 --> 00:21:00,230 analyze and identify items in photos. 330 00:21:00,240 --> 00:21:01,470 Very very cool. 331 00:21:01,470 --> 00:21:02,250 Awesome work guys. 332 00:21:02,250 --> 00:21:03,710 We'll see in the next video.