Identification Of Semi-Solid Liquids Using Photodiode And RGB Sensor With Spectrophotometric Neural Network (S-NN) Method

Droplet or human saliva is a semi-solid liquid that Covid-19 can catch on its patch media. It is also one of the causes of the fastest spread of Covid-19, resulting in a pandemic nowadays. So in this preliminary study, we created a tool that uses spectrophotometry to identify semi-solid liquids, including saliva, yogurt, and yeast water. The non-monochromatic spectrophotometric output will be classified using the neural network (NN) method. NN identifies the type of liquid by calculating the weight of each absorption wavelength of each semi-solid liquid sample from a non-monochromatic spectrophotometer. This initial research reveals several types of wavelength spectrum that can be recognized by Photodiode and RGB sensors through non-monochrome spectrophotometric methods. From the test results, saliva samples on glass media have a very high error rate of 99.9098%. For the overall average of saliva samples in all media, the accuracy is 89.1036%, and the error is 10.8964%. For the yogurt sample, the accuracy is 99.3075%, and the error is 0.6925%. The accuracy of the media without liquid is 78.8809%, and the error is 21.1191%. Based on the results, we found that the device can work properly as its aims.