Hi. I'm currently developing an app for android which could detect and recognize images of leaf diseases based on their symptoms like blights, lesions and etc.. As of the moment, I'm doing it with ORB and Bruteforce for feature detection as well as matching. One thing though, the accuracy seems an issue. Sometimes, an input image wrongly matched with the template images.
Is there any way to improve the accuracy using the following method? Or is there any other way? I though Neural Network would be the best option but I don't know how to implement it especially with Android. Please guys. Help me with these. I have 4 days left before the submission of the app.