Plant identification is a very well defined problem with masses of existing, well catalogued, information, that can be used to develop and test the Machine Learning models.
In my experience, data in industrial assets is not in the same good shape. For sure there’s masses of it but it’s not well catalogued and the problems are not clearly defined. A photo of a leaf is a completely self-contained problem statement*; you don’t then also need to enter the time of day or the season of the year or the soil type or the temperatures.
One day we may have such tools for industrial data applications but at this point I firmly believe we have a lot of data cataloguing to do before we can get there. We need some modern day “data Darwin’s” to go through all the data identifying features an cataloguing them for future reference. I think the technical name would be “taxonomists” (as opposed to “taxidermists” who make powerpoints – they look nice but the data is dead!). Ok, enough of my niche jokes! I’d love to hear what other people think ?.
*the plant has to have leaves in order for the app to work so it doesn’t work in the periods of a plants life when it has no leaves.