00:07Hey guys, in today's video, we are going to build a really useful project, a waist-shorting
00:12system using edge-emphas and Arduino Uno Cube.
00:15So what does it do?
00:16You just take a piece of waist like this and place it in front of a camera and the system
00:20will automatically identify whether it's biodegradable or non-biodegradable and short it for you.
00:26No manual effort, no confusion, everything happens automatically.
00:29Sounds helpful, right?
00:30Now, I will guide you step by step from collecting the data, training the model and finally making
00:37your hardware to respond like a real smart system.
00:40Let's get started.
00:46Now I am going to show you how to create an object detection model using edge-emphas and
00:50reply it to your Arduino Uno Cubeboard.
00:52First, open your browser and search for edge-emphas.
00:56Go to the website and login to your account.
00:58If you don't have an account, simply register and create a new one.
01:04Once you logged in, go to your profile and click on Create New Project.
01:08Give your project a name based on your use case.
01:20Now click on Collect New Data.
01:22Here you have two options.
01:24You can either scan the QR code using your mobile and start collecting data or you can upload,
01:29download data sets from your previous projects.
01:35In our previous object detection using edge-emphas, we have already mentioned how to collect the
01:40data and use it.
01:41Take a look at those to get a clear idea.
01:44I already have the collected data from the previous project so I am using it for now.
01:55After collecting the data, the next step is labeling.
01:58Make sure you properly label each image because this directly affects your model accuracy.
02:03Once labeling is done, go to the left side menu and click on Create Impulse.
02:07Inside Impulse 1, click on Add Processing Block and select Image.
02:14Then click on Add a Learning Block and choose Object Detection.
02:20After that, click on Save Impulse.
02:22Now you will see a success message.
02:24Next, click on the Image section just below the Create Impulse option.
02:32Here click on Save Parameters.
02:34After saving parameters, click on Generate Futures.
02:37Within a few seconds, your futures will be generated.
02:44Now move to the Object Detection section from the left side.
02:48Leave all the settings as default and click on Save & Train.
02:52The training process will take a few minutes.
02:59Once it's done, you will see the model's performance.
03:02If your accuracy is above 75%, that's great.
03:05If it's below 75%, you need to collect more data and train again.
03:11Finally, go to the Deployment section on the left side.
03:13Select your deployment target as Arduino UnoCube and click on Build.
03:23After a few minutes, your model will be downloaded and ready to use on your hardware.
03:30Yeah!
03:31The HM Burst part is complete.
03:34Now minimize the Browser tab and connect all the components as shown in the circuit diagram.
03:39Next, open Arduino App Lab.
03:43Before starting, make sure you complete all the required configuration in both software and
03:48your Arduino UnoCube.
03:49Once everything is set, go to the Example section and search for Detect Objects Using Camera.
03:55Open that example and take a moment to understand how it works.
04:00After that, click on the Copy & Edit option.
04:03Available at the top right corner and give it a name.
04:09Now go to the My App section where your copied app will be available.
04:17Inside the My App section, click on Create New App and give a name for your application.
04:23You can give any name you want like Bin, Wayshorting Project or Wayshorting Reganization Project,
04:30anything you want.
04:40Now you can see, the app you created is available in My App section.
04:45Next open the CLI in Arduino App Lab and enter your password.
04:50Then type the following command.
05:14After executing this, go back to the Example section and then return to your newly created
05:18app.
05:19Not the app you copied from the Example section.
05:22The app you created in the App section by clicking Create New App, which is you named
05:27as Bin, Wayshorting or Wayshorting Project like that.
05:31Inside your app, you will see the Asserts folder now, which is not previously present.
05:36Now at the top, click on Add Skits Library and include the required libraries like Servo
05:41and Arduino Router Bridge.
05:56Next click on Add Brick section.
06:02From that, add the WebUI HTML Brick and Video Object Detection Brick.
06:19After adding, click again on the Video Object Detection Brick.
06:23Inside that, go to the AI Models section and select Train a New Model.
06:29Click OK and connect it with Edge Impulse.
06:37Once redirected, open your recently created project in Edge Impulse and upload your trained
06:43model.
06:59Then come back to the Arduino App Lab.
07:02In the AI Models section, you will now see your model.
07:05Click on it and install the model.
07:13After installation, exit the app and re-enter it again.
07:17Now select your model in AI Models section as the active model.
07:24Next, open the Python file and paste the code from the Python folder.
07:31Similarly, open the Sketch file and paste the required Arduino code.
07:43Once everything is set, click on Run.
07:50This is the prototype for shorting and the camera is mounted here to detect which is what
07:55object object like Gandalf and the motor is mounted here.
07:58So, it can able to tilt which is biodegradable or non-biodegradable for the waste shorting.
08:03This is the hub which is used in the system, which is used to connect all components like
08:07Arduino UnoQ, the system as well as the camera.
08:11As you can see, the buzzer is connected to the 8th pin of the Arduino UnoQ.
08:15The motor is connected to the Arduino UnoQ's 9th pin.
08:18As you see, this is the camera portion where we can able to see whether the object is
08:23correctly recognized or not.
08:27If you take a battery like this and place it in front of the camera, the buzzer will get
08:34activated for the hazardous waste-like battery.
08:36The next thing, if you take a plastic waste like this, place it in front of the camera,
08:41the system will identify whether it is biodegradable or non-biodegradable.
08:45Ok, let me move on to next.
08:47When we place a waste of like this, it is biodegradable, right?
08:51In the front of the camera, the system will identify whether it is biodegradable or non-biodegradable
08:57on shorted for you.
08:58If you feel the project is more useful, hit the like button and subscribe to Succute Edges
09:02for more such videos.
Comments