1 00:00:00,580 --> 00:00:01,150 Hello. 2 00:00:01,210 --> 00:00:02,140 Welcome back. 3 00:00:02,150 --> 00:00:05,410 And this lesson I'm going to give you a road map of the course. 4 00:00:05,410 --> 00:00:11,770 In other words we're going to see the um the steps and the sequence of steps that we shall take in order 5 00:00:11,800 --> 00:00:17,960 to be able to deploy state of the art models on our microcontroller. 6 00:00:18,070 --> 00:00:24,080 So the course the first practical section is about the building blocks of neural network. 7 00:00:24,250 --> 00:00:29,980 And in this section the court is going to be written in C language just to show you how we can build 8 00:00:30,040 --> 00:00:35,950 simple neural network into the various building blocks from a C language perspective. 9 00:00:35,950 --> 00:00:39,640 Then we are going to be building complete modules over here. 10 00:00:39,640 --> 00:00:47,140 We're going to use Python cause if you're building neural networks in the real world there is a 99 percent 11 00:00:47,170 --> 00:00:49,390 chance that you're going to be using Python. 12 00:00:49,390 --> 00:00:53,690 So over here we'll see how to build our models completely in Python. 13 00:00:53,740 --> 00:00:58,750 So I've added a section for those of you who are not familiar with Python. 14 00:00:58,750 --> 00:01:05,050 I have other Python courses I've taken lessons from that course known as getting started with Python 15 00:01:05,170 --> 00:01:07,060 or Python essentials. 16 00:01:07,060 --> 00:01:12,370 And if you're new to Python when you complete that section you should be able to write Python code. 17 00:01:12,400 --> 00:01:19,510 So we're going to build complete modules in Python then we going to use other libraries popular libraries 18 00:01:19,960 --> 00:01:22,210 such as Caris and tensor flu. 19 00:01:22,270 --> 00:01:27,560 We're going to be built in modules which carries intensive flu because in the real world when you're 20 00:01:27,580 --> 00:01:31,450 building models you are going to use these libraries as well. 21 00:01:31,600 --> 00:01:38,410 Once that is done we are going to see how to quantized the models that we build use in terms of flu 22 00:01:39,910 --> 00:01:48,400 use in terms of flu for our embedded device in order to be able to deploy our tensor flu and care US 23 00:01:48,400 --> 00:01:48,870 models. 24 00:01:48,880 --> 00:01:53,470 We need to quantized them so that they can fit into the memory of our microcontroller. 25 00:01:53,470 --> 00:01:55,260 We're going to see how to do that. 26 00:01:55,330 --> 00:02:01,630 So that's going to be one step we would take and then we'll see how to deploy tens of flu models on 27 00:02:01,630 --> 00:02:08,850 our microcontroller and over here we're going to be deploying the models and see could we we build the 28 00:02:08,860 --> 00:02:15,850 neural network in Python and then once the neural network is built and we train it we get a weight with 29 00:02:15,850 --> 00:02:22,060 quantized a weight and then we come to a microcontroller and we write C code to perform what is known 30 00:02:22,060 --> 00:02:29,020 as inference so we deploy our tensor flow models on our our microcontroller and over here we're going 31 00:02:29,020 --> 00:02:31,520 to be applying C code. 32 00:02:31,620 --> 00:02:35,300 Next we'll be deploying models using in DeKalb mixed to a. 33 00:02:35,500 --> 00:02:42,950 This is a library built by SD microelectronics for machine learning on SD microelectronics devices. 34 00:02:43,030 --> 00:02:48,490 So we're going to see how to be deploying models using these uh this library as well. 35 00:02:48,490 --> 00:02:52,780 And then we're going to be building models using the coffee framework. 36 00:02:52,900 --> 00:02:58,800 The reason we'll be using the coffee is that the cafés the the framework best suited for use in these 37 00:02:59,230 --> 00:03:06,280 arm seems this new network library so arm has a library known as CMBS and then in that it was seems 38 00:03:06,280 --> 00:03:13,770 this neural network in the models that the library works perfectly with all the models. 39 00:03:14,050 --> 00:03:19,690 The framework that a library is easily workable with is the coffee framework which is a very simple 40 00:03:19,690 --> 00:03:21,580 framework for building your network. 41 00:03:21,620 --> 00:03:29,800 We're gonna get acquainted with coffee and build someone else there and then cry out with the wood it 42 00:03:29,800 --> 00:03:36,400 seems is and then and of course after we've built up the models in coffee we going to quantized them 43 00:03:36,660 --> 00:03:42,580 and to quantized we're going to write a python script to quantized these loose so that we'll be able 44 00:03:42,580 --> 00:03:46,380 to fit them into the memory of our microcontroller. 45 00:03:46,510 --> 00:03:50,680 So we'll be performing quantized ation of our coffee models as well. 46 00:03:50,680 --> 00:03:57,040 And then once that is done we'll see how to deploy our coffee models with the CMC as an end library. 47 00:03:57,040 --> 00:03:59,020 And this is going to be written in C code. 48 00:03:59,680 --> 00:04:01,570 So this is the um the roadmap. 49 00:04:01,630 --> 00:04:07,840 If you've purchased the course and if you don't find any of these topics in this section Do not be. 50 00:04:07,840 --> 00:04:12,310 Don't be bothered or do not worry at the end of the course when the course is complete. 51 00:04:12,340 --> 00:04:13,570 All of this should be there. 52 00:04:13,570 --> 00:04:20,140 So if you've purchased it now and you don't find topics relating to let's say coffee or terms of flow 53 00:04:20,470 --> 00:04:22,640 then they are still in production. 54 00:04:22,690 --> 00:04:29,010 This is the complete picture of the course and you can hold me to task on this right. 55 00:04:29,050 --> 00:04:31,210 So yeah if you have any questions just let me know. 56 00:04:31,240 --> 00:04:33,370 And this is the roadmap. 57 00:04:33,370 --> 00:04:34,570 I'll see you in the next lesson.