1 00:00:00,510 --> 00:00:07,140 So let's move forward and visualize the data we have, so here I'm visualizing some of the traffic signs 2 00:00:07,140 --> 00:00:07,680 from the data. 3 00:00:09,250 --> 00:00:11,100 And let's see how it looks. 4 00:00:14,460 --> 00:00:16,890 So here are some of the images from the dataset. 5 00:00:17,810 --> 00:00:23,860 And as you can see here, the dimension of the images are not seen for all the images. 6 00:00:23,880 --> 00:00:28,220 So as the size of the images are different, we have to make them equal. 7 00:00:28,370 --> 00:00:35,420 So we will take the mean of these images here and finding out the mean of these images so we can we 8 00:00:35,420 --> 00:00:40,060 already know that there are forty three classes in this graphic, same dataset. 9 00:00:40,430 --> 00:00:43,090 So I'm running a for loop from zero to 43. 10 00:00:43,100 --> 00:00:44,270 That is what it is. 11 00:00:44,990 --> 00:00:48,040 And I'll take each image and find out the meaning of that image. 12 00:00:48,090 --> 00:00:51,140 I mentioned one and I mentioned to and will store it. 13 00:00:52,840 --> 00:00:57,730 Let's run this and store the dimension of these images to find out the meaning of these images. 14 00:00:58,090 --> 00:01:01,150 It's running right now, so we have stored this. 15 00:01:01,570 --> 00:01:05,670 Next, we will find out the meaning of these dimensions and print them out. 16 00:01:07,790 --> 00:01:16,430 So here we can see that 50 by 50 is the average for all the images here, we can see that the damage 17 00:01:16,430 --> 00:01:20,720 and one win is fifty point three two and damage in two minutes, fifty point three. 18 00:01:20,990 --> 00:01:23,840 So we can conclude that 50 50 is the average. 19 00:01:24,350 --> 00:01:28,340 Now, we will convert all these images into the ship, 50 by 50. 20 00:01:30,330 --> 00:01:35,510 For that, I'm using images, an empty list also unlabelled is an empty list. 21 00:01:37,400 --> 00:01:45,230 And we ship all these images and also store their label ideas for this, we have we are running a loop 22 00:01:45,230 --> 00:01:49,960 here for forty three classes and storing all the images, resizing them. 23 00:01:49,970 --> 00:01:51,890 And also there is also. 24 00:01:52,940 --> 00:01:59,970 Mr. Anderson, and convert these images or resize these images into the desired dimensions. 25 00:01:59,990 --> 00:02:03,040 That is 50 by 50 and also storing the. 26 00:02:06,200 --> 00:02:12,890 It's done so the next step would be to convert these images into no error and also to normalize them. 27 00:02:12,890 --> 00:02:19,070 So in order to normalize an image, we will divide it by two fifty five because the value of each image 28 00:02:19,070 --> 00:02:20,700 ranges from zero to 255. 29 00:02:20,930 --> 00:02:25,730 So we will divide each image by two fifty five to normalize them and to convert them into array will 30 00:02:25,730 --> 00:02:26,940 simply use the function. 31 00:02:26,960 --> 00:02:32,960 And so let's Grandison and convert their images into numbers and normalize them. 32 00:02:35,540 --> 00:02:36,230 So it's done. 33 00:02:38,050 --> 00:02:42,760 Similarly will store the label I listened to today and will see the shape of it. 34 00:02:45,210 --> 00:02:52,260 So here we can observe that there are three nine two zero nine ladies, similarly will print the shape 35 00:02:52,260 --> 00:02:53,130 of the images. 36 00:02:55,330 --> 00:03:02,490 So here we can see that there are three nine two zero nine images with the shape of fifty to fifty three. 37 00:03:03,010 --> 00:03:07,900 So her three is the channel, which means that they are colored images in AGP format. 38 00:03:07,930 --> 00:03:09,220 Let's move forward now. 39 00:03:09,220 --> 00:03:13,900 We'll visualize the number of the class to find out if the data is balanced or not. 40 00:03:13,960 --> 00:03:19,740 So here we can see that data is around twenty two hundred in each class. 41 00:03:19,750 --> 00:03:23,380 So we can say that the data is not imbalanced, data is balanced. 42 00:03:23,390 --> 00:03:25,190 We do not need to balance the data. 43 00:03:25,210 --> 00:03:26,500 Data is already balanced. 44 00:03:27,510 --> 00:03:30,210 Now we'll split that data using the test transmit. 45 00:03:33,060 --> 00:03:38,040 Into a ratio of 80 percent to ending data and 20 percent validation later. 46 00:03:40,980 --> 00:03:48,600 Similarly will convert the training data that will convert the label, that is which class it belongs 47 00:03:48,600 --> 00:03:52,570 to and do not including using too categorical function. 48 00:03:53,220 --> 00:03:53,910 So why not? 49 00:03:53,910 --> 00:03:58,910 Encoding is important because if we don't apply, we're not including the machine will prioritize it. 50 00:03:58,920 --> 00:04:03,530 So to avoid prioritization, we apply, but not encoding, so will apply. 51 00:04:03,540 --> 00:04:09,300 We're not encoding on the training and the validation data, the target variable that is in categorical 52 00:04:09,300 --> 00:04:10,500 and well, categorical.