1 00:00:00,080 --> 00:00:02,900 Now we understand the problem we are trying to solve. 2 00:00:02,930 --> 00:00:05,390 We'll go ahead and download our data set. 3 00:00:05,390 --> 00:00:11,870 So first things first, we'll head over to Kaggle where we're going to make use of Rajkumar's data set 4 00:00:11,870 --> 00:00:15,020 entitled The People Coding segmentation. 5 00:00:15,230 --> 00:00:16,010 Um, data set. 6 00:00:16,010 --> 00:00:21,590 That's all the images of people and semantic segmentation, masks of their different codings. 7 00:00:22,010 --> 00:00:29,540 Um, if you scroll down here, you'll see this Jpeg images and the corresponding Jpeg masks, same as 8 00:00:29,540 --> 00:00:33,050 the PNG images and the corresponding PNG masks. 9 00:00:33,050 --> 00:00:39,950 So this is the data set we are going to be working with, which also comes with this labels csv file. 10 00:00:39,950 --> 00:00:48,080 In this label CSV file we have all the different classes which we are going to use in our data set. 11 00:00:48,080 --> 00:00:54,080 So we go from accessories right up to uh wedges. 12 00:00:54,080 --> 00:00:57,740 So we have 59 classes including this null. 13 00:00:57,740 --> 00:01:00,650 That is um, to say we have the background. 14 00:01:00,650 --> 00:01:08,570 Now if we scroll back up here and we open this up, you find that for every image, as we had said already, 15 00:01:08,570 --> 00:01:11,180 we are going to have a corresponding mask. 16 00:01:11,180 --> 00:01:14,150 So this is image 0001. 17 00:01:14,150 --> 00:01:17,390 If you open this up you have this image of this lady. 18 00:01:17,390 --> 00:01:21,560 And then we use this and check out this corresponding mask. 19 00:01:21,560 --> 00:01:23,300 So open this up. 20 00:01:23,300 --> 00:01:25,340 We have there we go. 21 00:01:25,340 --> 00:01:26,960 We have its corresponding mask. 22 00:01:26,960 --> 00:01:33,440 In summary, for every pixel point in the image there is a corresponding point in the mask. 23 00:01:33,440 --> 00:01:35,390 And there's different images. 24 00:01:35,390 --> 00:01:40,970 And masks are stored in different folders in our data set. 25 00:01:40,970 --> 00:01:48,140 Before we proceed to download this data set, we should note that it contains a thousand images and 26 00:01:48,140 --> 00:01:56,300 segmentation masks, pairs of individual people's clothings, and 59 object classes, which we had seen 27 00:01:56,300 --> 00:01:59,180 already in the labels CSV file. 28 00:01:59,180 --> 00:02:03,920 Then also we could go ahead and click on settings. 29 00:02:04,680 --> 00:02:08,970 We scroll down and then we click on Create New Token. 30 00:02:09,000 --> 00:02:13,530 Once you click on Create New token, you will have a Kaggle dot Json file. 31 00:02:13,530 --> 00:02:20,880 And once you have this Kaggle dot Json file downloaded, you then go ahead and upload it to Google Colab. 32 00:02:20,880 --> 00:02:23,430 So this is our Kaggle dot Json file right here. 33 00:02:23,430 --> 00:02:26,310 And now we install Kaggle. 34 00:02:26,310 --> 00:02:29,070 We are going to make this new directory. 35 00:02:29,070 --> 00:02:33,630 We're going to copy our Kaggle dot Json file into our Kaggle directory. 36 00:02:33,630 --> 00:02:37,500 Then we'll finally download our dataset. 37 00:02:37,500 --> 00:02:38,970 Let's get back. 38 00:02:39,240 --> 00:02:40,680 Um let's go back. 39 00:02:40,680 --> 00:02:44,250 And then we copy this command. 40 00:02:44,250 --> 00:02:47,430 So here we have let's reduce this so you could see it clearly. 41 00:02:47,430 --> 00:02:50,580 So we have copy API command. 42 00:02:50,580 --> 00:02:55,470 So if you copy this and then you paste our leisure spaces out down here. 43 00:02:55,620 --> 00:03:00,000 If you paste this out here you see you have Kaggle datasets download. 44 00:03:00,000 --> 00:03:04,350 And then you have the the owner that's the creator of this data set. 45 00:03:04,350 --> 00:03:07,770 And then you have the data set name. 46 00:03:07,770 --> 00:03:08,910 So that's it. 47 00:03:08,910 --> 00:03:11,040 We run this cell. 48 00:03:11,040 --> 00:03:18,270 And then given that the output or what we are going to download is going to be this zip file, we are 49 00:03:18,270 --> 00:03:22,200 going to unzip this into our data set folder. 50 00:03:22,800 --> 00:03:28,140 After the downloading and unzipping is done we open up this files. 51 00:03:28,140 --> 00:03:33,570 And then you could see that we have this people clothing segmentation dot zip file. 52 00:03:33,570 --> 00:03:42,300 And then we have in our data set folder we have exact same folders we had seen already in this Kaggle 53 00:03:42,300 --> 00:03:44,820 data set on the Kaggle platform itself. 54 00:03:44,820 --> 00:03:46,950 So right here we have images. 55 00:03:46,950 --> 00:03:48,840 We have the masks. 56 00:03:48,840 --> 00:03:53,400 And then we could now prepare this data set.