I'm working on developing an Android app that can recognize a leaf's species by photographing it. I heard about the Tensorflow API and wanted to retrain the model with my Dataset. I tried different optimizations like :
Increasing the training steps
Using random cropping/scaling/brightnessing functions
Still, my model isn't accurate and confuses all the leaves. I guess it's because there is a lot of leaves that are very similar in shape. I searched a lot on the web and can't find a good way to fix that lack of accuracy.
Is Tensorflow adapted to my problem, or should I use another method ?