1
00:00:00,810 --> 00:00:07,590
Lesson to all the student tragic in this course, we have already learned how to use Edge Impulse and

2
00:00:07,590 --> 00:00:14,380
TensorFlow Lite for microcontrollers to train and deploy highly optimized models for inference with

3
00:00:14,400 --> 00:00:15,960
real terminology, right?

4
00:00:16,860 --> 00:00:22,650
You also should have understanding of how to combine model inference with integration to cloud services

5
00:00:23,190 --> 00:00:25,800
and graphical user interface.

6
00:00:26,580 --> 00:00:31,890
But of course, learning and making doesn't stop here for the last lesson of the course.

7
00:00:32,220 --> 00:00:38,700
We encourage you to think outside of the box and create a project of your own number of projects with

8
00:00:38,700 --> 00:00:46,080
tiny email, a growing day by day and below are some of the most interesting ones our team found on

9
00:00:46,080 --> 00:00:46,680
the internet.

10
00:00:47,340 --> 00:00:53,840
Here we have community projects, dynamical water sensor based on edge impulse and Arduino Sense, though

11
00:00:53,850 --> 00:00:55,940
that's amazing right now.

12
00:00:56,190 --> 00:00:58,200
You can find this one on this link.

13
00:00:58,800 --> 00:01:06,780
Now this is actually an implementation to identify running water, tap sound and once heard a buzzer.

14
00:01:07,170 --> 00:01:09,780
Plus a LED timer is triggering.

15
00:01:10,500 --> 00:01:16,920
The device is listening continuously for any sound, and once a running tape is identified, a timer

16
00:01:16,920 --> 00:01:21,990
is triggered to produce 20 seconds buzzing sound and a flashing LED light.

17
00:01:22,560 --> 00:01:29,890
The project is based on a two components of Arduino, Nanotech, three Belizeans and edge impulse to

18
00:01:30,180 --> 00:01:31,830
as a developing platform.

19
00:01:32,160 --> 00:01:33,600
Now that sounds amazing, right?

20
00:01:34,350 --> 00:01:41,160
Another is another community projects is a tiny amount keyword detection for controlling IGB lights.

21
00:01:41,190 --> 00:01:42,810
Now that's interesting.

22
00:01:43,860 --> 00:01:50,700
Do you somehow train a TensorFlow model to recognize certain key routes and control an RGV light strip

23
00:01:50,700 --> 00:01:54,420
using and Arduino Nano 33 a license?

24
00:01:55,380 --> 00:01:59,820
Here is an easy G Analyzer powered by edge impulse.

25
00:01:59,850 --> 00:02:02,550
Now that's interesting for the medical field.

26
00:02:03,120 --> 00:02:10,830
Now this works a tiny based medical device powered by edge impulse to predict atrial fibrillation,

27
00:02:11,310 --> 00:02:18,840
atrial ventricular block one and a normal EKG with more than 90 percent accuracy.

28
00:02:19,710 --> 00:02:27,330
And also, it is an application powered by Edge Impos to develop a mini diagnosis EKG analyzer device,

29
00:02:27,750 --> 00:02:29,220
which can fit in a pocket.

30
00:02:29,620 --> 00:02:36,570
Now that's convenient, and it can diagnose heart diseases independently without a cloud connectivity.

31
00:02:36,810 --> 00:02:41,370
Now that is very, very convenient for the medical people.

32
00:02:41,820 --> 00:02:49,230
Another medical device that was invented is a fall detection and heart rate monitoring using an AVR

33
00:02:49,650 --> 00:02:50,410
aortic.

34
00:02:50,850 --> 00:02:56,760
Now this is a wearable low power device that detects falls using machine learning at the edge.

35
00:02:57,270 --> 00:03:03,720
It also monitors the heart rate full scan result in physical and physiological trauma, especially for

36
00:03:04,110 --> 00:03:05,310
elderly patients.

37
00:03:05,340 --> 00:03:13,410
Now, in order to improve the quality of life of these patients, this project presents the development

38
00:03:13,410 --> 00:03:21,990
of a fall detection and heart rate monitoring system, so this detects any fall in the heart rates of

39
00:03:22,140 --> 00:03:22,860
the patients.

40
00:03:23,310 --> 00:03:29,430
The wearable device obtains information from accelerometer and heart rate sensor and sends them to the

41
00:03:29,820 --> 00:03:31,560
A.W. security guard.

42
00:03:31,980 --> 00:03:36,840
The fall detection is done offline on the chip, using a machine learning algorithm.

43
00:03:37,410 --> 00:03:41,790
How convenient this device is because it can run even though offline.

44
00:03:42,120 --> 00:03:46,360
Here is a build a smart dumbbell with some detail.

45
00:03:46,360 --> 00:03:46,680
21.

46
00:03:46,680 --> 00:03:54,950
Machine learning every key now distance your dumbbell into a smart belt with the microchip some 21 MLF

47
00:03:55,270 --> 00:03:59,730
and an embedded classifier built in the edge imposter.

48
00:03:59,730 --> 00:04:00,150
VIDEO.

49
00:04:00,660 --> 00:04:04,110
Here are some ideas from the seed Edu team.

50
00:04:05,160 --> 00:04:07,500
One is a better sound translator.

51
00:04:08,070 --> 00:04:15,300
It actually uses real terminal internal microphone to record the data sets of sounds of your pet being

52
00:04:15,300 --> 00:04:23,700
hungry, angry, lonely, happy, excited or anything, and then the train model that would help you

53
00:04:24,240 --> 00:04:25,740
to understand it better.

54
00:04:26,460 --> 00:04:35,370
Here we have a very small project, which is a robot control with EMG detector that that's useful.

55
00:04:36,060 --> 00:04:43,800
It utilized Grove EMG detector module to analyze arm muscle activity and control your own robot with

56
00:04:43,800 --> 00:04:44,880
arm gestures.

57
00:04:44,950 --> 00:04:46,410
Now that's fantastic, isn't it?

58
00:04:47,220 --> 00:04:51,840
Another idea from the seed Edu team is an intelligent plant care.

59
00:04:52,500 --> 00:04:59,490
Now this employed to rate the temperature and humidity and pressure, we determine the internal light

60
00:04:59,490 --> 00:04:59,910
sensor.

61
00:04:59,980 --> 00:05:06,040
And possibly grove moisture north to monitor and control plant environment for optimal growth.

62
00:05:06,700 --> 00:05:12,790
So basically what this does is it monitors the growth of a plant.

63
00:05:13,990 --> 00:05:18,670
Here we have is an advanced technology, a real terminal on beetle.

64
00:05:18,880 --> 00:05:21,310
So as we all know that this is a robot.

65
00:05:22,210 --> 00:05:27,360
It installs real terminal on Bionic Robotic Dog from Pito LLC.

66
00:05:27,380 --> 00:05:36,280
That and seed add to both beetle and real terminal have Raspberry Pi pin connectors, which makes interfacing

67
00:05:36,280 --> 00:05:37,210
them easy.

68
00:05:38,660 --> 00:05:40,430
That's it for this presentation.

69
00:05:40,580 --> 00:05:44,900
I hope you've learned a lot, and I'm looking forward for your future projects.

70
00:05:45,590 --> 00:05:54,350
If you have any questions, feel free to contact me on Skype, on my phone number and on my email once

71
00:05:54,350 --> 00:05:54,680
again.

72
00:05:55,130 --> 00:05:56,180
Thank you for listening.