domingo, 28 de novembro de 2021

How to predict concentrations using smartphone camera



Colorimeter: Wine in water example

Getting started

1. Collecting and saving samples

First of all, you will need to grab some colors, in our example we have collected 3 samples of wine in different percentages in pure water. We also have standardized the environment light, positioned the samples in front of our target and clicked in the bottom button to save. We also have attributed a name for each sample to quick found in the database later.
Sample name "1": 10% of wine in water.

Sample name "5": 50% of wine in water.

Sample name "111": 100% wine.

2. Creating one application using Creator Studio

Once we have a few samples in our database, we can navigate to Create application, and there we can click in 'plus' button to start application creation. Type a name, in our example is Wine %.


After created, click in your application name, and you will have 3 new tab's. Database, Inputs and Train. In Database tab you can search for the samples that you have saved before. Click on each of them one by one and insert the reference/known value (Y) that you want to predict in the future.


Once you have finished the reference values attribution, in Inputs you will be able to see all of your selected samples. Then just go to Train, select your desired X input type (Hue,R,G or B) and click Train Model button. Now you will be able to see the results, then just click save using top button in toolbar.


3. Using your application in real time

To use your new application, go to menu and navigate to Use in Real Time, and start measuring your new samples. Here the model will make the predictions automatically and will produce a blue line of each predicted value. You also can set manually a upper and lower limit if you want.

In the example bellow, we can see our model make predictions of each wine % in water in real time!!!


#android #googleplay #color #colorscience #laboratory #lab #chemistry #education #smartphone #machinelearning #datascience #chemometrics #statistics #math #labtools #tools #science #innovation #research #industry #qualitycontrol #processcontrol

Nenhum comentário:

Postar um comentário