Simon Sayz/Nice House of Plastics Chairs Classifier

Created Fri, 21 May 2021 00:00:00 +0000

Background of App

So i’ve been working as a Business Systems Analyst at Nice House of Plastics for a few months now. One of my smallest challenges was to memorize the names of Our Best Selling Chairs. i.e All Size Vs Portable. It’s not a big problem but why not?

Solution 💽

I wanted my Solution to be Simple & Fast. So i decided to work on a machine learning model and serve it through a Mobile Application. So i could always take a pic with my Phone’s Camera and get the Prediction.

How ? ☄️

To train a model, you need a lot of training data, say 1000 for each chair. The obvious method would have been to take a lot of pictures on my phone and use them as training data. After taking about 10 Pics, i realized there has to be a better way:-

  • Record a 2 Min Video of each Chair on My Phone.
  • Open the Video in Photoshop and Export each Frame of the Video as an Image
  • Use tensor flow the modeling.
  • Flutter for the Mobile Application.

Anhaa!! 🤔

The Question still remains?? Does it work?

Conclusion !

Am no expert on Machine Learning and i don’t think this is the best way to do it, but it works.

You can also get yourself One of Our Chairs and Test this Application.

What Next

Am thinking of training a model that can do predictions on all Our Products and may be a mini recommendation engine on Our Website.