I have been wanted to increase the number of supported objects for the active vision system for the prosthesis, as it means that there will be more supported grip patterns corresponding with the objects. Here is my weak attempt of creating a banana classifier.