This is a project I recently did called "Project Sneaky Peanut Butter" wherein I gathered 7 kinds of Peanut Butter, A RedBoard (Arduino), and a Spectral Triad Sensor https://learn.sparkfun.com/tutorials/spectral-triad-as7265x-hookup-guide?_ga=2.230455772.1578243289.1643145700-1853259630.1615791300
to gather spectral signatures across 18-channels from peanut butter samples, in an attempt to classify/detect them with a Keras Model...just to see if it was possible
Here is the data https://docs.google.com/spreadsheets/d/1_E5vRQhQIjnmQlrrsLSmfgA23quHl1xvFlmkLIFkn8g/edit?usp=sharing ...and a legend for the wavelengths of the channels AS7265x.pdf
Not a strictly useful project, but I can run the ML model in real-time with an unknown sample and tell you which style of PB it is :-D (If you spend some time simply exploring the data first-hand, a few things immediately stand out...Simply JIF has the highest R channel peak avaerage, the "shape" of the A, F, R, and W peaks help delineate many out-right)
If I were to form the same "model" into speech/an algorithm, I'd say: Is there 3000+ R-channel?
-> YES? It's either Simply JIF, JIF Honey, or Kroger Honey
-
Multiple R values > 3250? = Simply JIF -
A > F = Kroger Honey -
A < F = JIF Honey
-> NO?
-
A > F = JIF Omega-3
....the last 2 are the hardest to distinguish, but Reduced Fat JIF has a larger W channel compared to its A and F channels, when compared to Kroger Creamy
NOTES: 1, the organic option, was disqualified on account of oil/foodstuff separation....so, 6 datasets