Thanks to modern chemistry, we are able to locate heaps of chemical compounds in water, even at extraordinarily low concentrations. Majority of techniques for compound analysis require state-of-the artwork lab facilities. Fortunately, we don’t have to check for everything! A smaller and extra sensible set of checks can offer excellent means of monitoring. Commonly test strips are dipped in water sample and compared by naked eyes to derive conclusions which can be harmful for the person dealing with contagious samples or may cause inaccuracy in results due to human error. An automated system to carry out these processes and eliminate the threat and inaccuracies is put in to works using deep learning. In our current work we have used Raspberry Pi integrated with a HD webcam to capture the strip images, extract RGB values and deduce the results. We address the application of Deep Learning to Raspberry Pi based calorimetric detection of water quality. The test strips are available from a general medical store and hence the database was created using the values provided beforehand . Once the database was created, it was trained using Neural Networks whose accuracy was turned out to be 95.11 %.
Manuscript: Automated Water Test Strip Reader
Watch the video to know more!