Class set required to do a fruit recognition system, in the BBS see based on color filtering, morphological operation, image processing operations such as edge detection to identify citrus, but this did not seem to involve the object characteristics, assuming that with an orange circle whether patterns can be identified as an orange? How to do it according to the object recognition features, improve the accuracy? Request to give directions!

Is to consider before go deep learning approach, object recognition model is almost finished, but stuck in places where the number of object recognition (also if you can, please show this), then go to image processing,