1 00:00:01,530 --> 00:00:07,530 ‫The input to our network when we are doing image recognition is an image. 2 00:00:09,030 --> 00:00:16,140 ‫Now the image can be either black and white, which is also known as grayscale image, or it can be 3 00:00:16,140 --> 00:00:19,860 ‫coloured in a grayscale image. 4 00:00:20,610 --> 00:00:28,410 ‫Each pixel contains only one information and the information is how wide is that pixel? 5 00:00:30,270 --> 00:00:34,550 ‫So usually it is seen on a scale of zero to 255. 6 00:00:35,850 --> 00:00:38,070 ‫255 means completely white. 7 00:00:38,940 --> 00:00:40,680 ‫Zero means completely black. 8 00:00:42,060 --> 00:00:46,270 ‫And any value in between zero and 255 is a shade of grey. 9 00:00:49,110 --> 00:00:54,780 ‫But in a colored image, each pixel has three values instead of one. 10 00:00:57,030 --> 00:01:03,690 ‫These three are R, G and B that is red, green and blue. 11 00:01:06,000 --> 00:01:13,890 ‫It turns out that receptors in human eye perceive color as the combination of these three colors only. 12 00:01:15,750 --> 00:01:17,700 ‫These are called primary colors. 13 00:01:19,530 --> 00:01:24,330 ‫If you mixed these three in different quantities, you will get different colors. 14 00:01:27,510 --> 00:01:35,040 ‫So each pixel has three values value for each of these primary color. 15 00:01:35,370 --> 00:01:37,740 ‫That is from zero to 255. 16 00:01:38,970 --> 00:01:45,330 ‫And depending on what combination of these three values do we have, we can see the corresponding color 17 00:01:45,450 --> 00:01:46,110 ‫and the pixel. 18 00:01:48,300 --> 00:01:52,980 ‫So if all three values are 255, we get white color. 19 00:01:55,350 --> 00:01:59,070 ‫If red and green are 255 and blue is 0. 20 00:01:59,730 --> 00:02:00,830 ‫We get yellow. 21 00:02:04,430 --> 00:02:11,930 ‫Here in this animation, you can see that when all three colors overlap, we get white collar. 22 00:02:13,140 --> 00:02:16,880 ‫When red and green overlap, we get yellow. 23 00:02:18,230 --> 00:02:24,690 ‫When blue and green overlap, we get to cian and so on. Here 24 00:02:25,390 --> 00:02:32,090 ‫I have also mentioned one example if you want to create purple color here are the values of our 25 00:02:32,090 --> 00:02:36,270 ‫RG and B setting these values will give you purple color. 26 00:02:39,560 --> 00:02:48,500 ‫So if each pixel has three values, we can say that the entire image is made up of three layers, one 27 00:02:48,500 --> 00:02:56,280 ‫layer made up of red color values called the red channel one layer made up green color values called 28 00:02:56,510 --> 00:03:02,690 ‫green channel, and one layer made up of blue color values called the Blue Channel. 29 00:03:04,340 --> 00:03:12,620 ‫So you need to remember that whenever we work with a grayscale dataset such as fashion amnest you 30 00:03:12,620 --> 00:03:13,880 ‫have only one channel. 31 00:03:15,170 --> 00:03:20,960 ‫But whenever we work on colored image, we need to take into account the three color channels 32 00:03:20,990 --> 00:03:21,410 ‫Also. 33 00:03:24,850 --> 00:03:29,650 ‫To demonstrate to you a color image is made of three channels. 34 00:03:30,490 --> 00:03:34,750 ‫I have created pixels using cells of Excel worksheet 35 00:03:37,170 --> 00:03:41,620 ‫continued that this pixel is a bundle of these three cells. 36 00:03:42,520 --> 00:03:46,150 ‫So the first pixel is bundle of three cells. 37 00:03:47,430 --> 00:03:53,840 ‫One cell to take red value, one will take green and one will be blue value. 38 00:03:55,820 --> 00:03:58,080 ‫And together they make one pixel 39 00:03:59,930 --> 00:04:05,000 ‫The value in these cells correspond to R, G and B values. 40 00:04:05,300 --> 00:04:10,310 ‫And we have used conditional formatting to color these cells depending on their value. 41 00:04:16,290 --> 00:04:21,370 ‫So if you look at this cell, this cell has value 231. 42 00:04:22,380 --> 00:04:24,420 ‫So it is giving us a bright green color. 43 00:04:26,220 --> 00:04:30,030 ‫The cell to the right of this has only 47 value. 44 00:04:31,080 --> 00:04:32,850 ‫Which is why it is dark. 45 00:04:33,930 --> 00:04:36,280 ‫So if it is zero, it will be completely black. 46 00:04:36,660 --> 00:04:39,240 ‫If it is 255, it would be bright green. 47 00:04:39,870 --> 00:04:44,340 ‫And anything in between is a gray scale of green color. 48 00:04:46,380 --> 00:04:49,170 ‫It similarly works with Red and Blue Channel 49 00:04:49,200 --> 00:04:49,620 ‫Also. 50 00:04:52,650 --> 00:04:55,410 ‫Now, if we are looking at individual pixels. 51 00:04:56,520 --> 00:05:01,290 ‫We cannot make out the image, but when we start zooming out. 52 00:05:04,460 --> 00:05:05,940 ‫Do you start seeing the image? 53 00:05:10,070 --> 00:05:13,580 ‫Do you see a normal colored image of the mario? 54 00:05:16,240 --> 00:05:17,500 ‫Isn't this somewhat magical? 55 00:05:18,580 --> 00:05:23,020 ‫We were able to create a colored image using only three colors. 56 00:05:24,460 --> 00:05:34,510 ‫You can see on the borders other than the mario, you can see white color, although there were only 57 00:05:34,510 --> 00:05:36,220 ‫red, green and blue colors. 58 00:05:37,550 --> 00:05:43,310 ‫This is because our eyes perceive combination of these three colors as white. 59 00:05:46,030 --> 00:05:53,020 ‫This is how our computer and mobile phone screens also wor for each point or pixel on the screen. 60 00:05:53,890 --> 00:05:59,950 ‫There are three small light bulbs, one for each red, green and blue color. 61 00:06:01,450 --> 00:06:07,480 ‫And with these three channels only, we see so many beautifully colored images on our screens. 62 00:06:11,910 --> 00:06:13,160 ‫I would like to mention here. 63 00:06:13,350 --> 00:06:16,320 ‫This Web site, which are used for creating this Excel image. 64 00:06:17,730 --> 00:06:19,430 ‫think-maths.co.uk 65 00:06:21,240 --> 00:06:24,510 ‫You can just upload any image here and it creates this 66 00:06:24,640 --> 00:06:27,000 ‫RGB Pixilated Excel file for it 67 00:06:28,170 --> 00:06:29,730 ‫It is an amazing and fun tool. 68 00:06:29,970 --> 00:06:30,660 ‫Do check it out.