Today we are going to teach you how to build a neural network using VIA Pixetto. This neural network is super simple and is a great introductory example of how AI machine learning works.
Alright, let’s get started!
Step 1
If not already available on your personal computer, please download VIA Pixetto Studio from www.pixetto.ai beforehand. In addition, connect your VIA Pixetto to your PC using a MicroUSB 2.0 cable.
(For more detailed instructions, please refer to our ‘Getting Started with VIA Pixetto Step-by-Step Guide’ ).
Step 2
Open the Pixetto Utility application in Pixetto Studio. Aim your VIA Pixetto lens at your object of choice.
In the lower right hand corner, click the ‘Record Video’ button to begin recording. Make sure that you move the lens around to capture all angles of the object. We should aim for a recording of 10 to 15 seconds.
Name the file and save it. Close Pixetto Utility once you have finished recording.
Step 3
Go to VIA Pixetto’s integrated online platform, www.pixetto.ai,and click on the ‘Tools & Docs’ tab in the upper right corner. Once on the ‘Documentation & Software Download’ webpage, scroll down to the very bottom.
Click the ‘Machine Learning Accelerator’ icon to access the Machine Learning Accelerator webpage. Once on the Machine Learning Accelerator homepage, click the ‘Machine Learning’ icon to access the ‘Machine Learning’ user interface.
Step 4
Enter your model name. Click the ‘Upload Video’ button to upload your pre-recorded video. Enter a one-word label name for your video that is preferably the same as your file name.
In the image which appears, draw a rectangle over your target object.
Wait for the application to process and play the video. Confirm whether or not your tracking result is correct.
Repeat the above steps to teach VIA Pixetto to track more objects. There is no limit to the number of objects tracked, however, the more objects you track, the longer the loading time will be. Make sure to upload 2-3 videos of objects for a more optimal processing period.
Step 5
Select ‘Popular Combination’ from the left-side column. Select your output class as the number of objects you have recorded. Click ‘Start’ in the upper right hand corner to start training.
VIA Pixetto will start training based off the object recordings you have “fed” it. While it is training you can see lines of code running down the Console Chart Diagram. Make sure Pixetto Utility is closed while training.
Step 6
Click ‘Download to Device’ in the upper right hand corner to download to VIA Pixetto.
Open Pixetto Utility. Aim your camera at the object recorded earlier.
In Pixetto Utility, select the ‘Neural Network’ function in the upper right hand corner and then click ‘Apply’ to start object recognition. VIA Pixetto will identify the displayed items and label them accordingly.
And we are done! That was easy, wasn’t it? While machine learning might sound like a complex idea, and it certainly can be, that doesn’t mean we can’t simplify it.
With the help of VIA Pixetto, we have successfully made machine learning more understandable and more fun.
Now, it’s your turn. Follow this tutorial and don’t forget to share your results with #VIAPixetto on your Facebook, Twitter, or Instagram for the chance to be featured on our official social media!
