Revision history [back]

I am doing a project to recognize one kind of leaf.I have 4000 positive images and 6000 negative images. My object have different shape.I am using emgucv visual studio 2010 c#. I want to use cascade classifier and then use cascade.xml in an embedded system to recognize leaves.I want to know what kind of features extraction is the best to use in my requirements? I read about haar is not the best in embedded system and mobile because haar use floating point calculations and I supposed those systems are not suitable for that calculations is that correct? but what about lbp or hog?

I am doing a project to recognize one kind of leaf.I have 4000 positive images and 6000 negative images. My object have different shape.I am using emgucv visual studio 2010 c#. I want to use cascade classifier and then use cascade.xml in an embedded system to recognize leaves.I want to know what kind of features extraction is the best to use in my requirements? I read about haar is not the best in embedded system and mobile because haar use floating point calculations and I supposed those systems are not suitable for that calculations is that correct? but what about lbp or hog?

I am doing a project to recognize one kind of leaf.I have 4000 positive images and 6000 negative images. My object have different shape.I am using emgucv visual studio 2010 c#. I want to use cascade classifier and then use cascade.xml in an embedded system to recognize leaves.I want to know what kind of features extraction is the best to use in my requirements? I read about haar is not the best in embedded system and mobile because haar use floating point calculations and I supposed those systems are not suitable for that calculations is that correct? but what about lbp or hog?