In order to enable our plant identification API to provide the best results, our tech stack uses Google’s Tensorflow and some python-based heavy lifting. Minor part of the service is based on Google Cloud and Microsoft Azure.
The neural network’s architecture is based on Inception model, but we have implemented a bunch of customization, mostly related to the certainty determination, adaptive represenative photo selection and improving the accuracy with some meta attributes like GPS coordinates.
Our newest model can identify 10.997 plants species. The set of species selection is based on the most identified plants within United States, European Union, India and Australia.
We are continualy testing the accuracy. It’s havily related on the quality of the photo, but the general success rate fluctuates between 80%-90%. In the future versions we’l stabilize the rate over 90% while widening the spectrum of classified species.