Susan J. Simmons, Devrim Unay, Karl Ricanek Jr., Bernard Gosselin (Belgium)
Random Forests, Support Vector Machines, pattern recognition, feature tuning, fruit inspection
This work compares two prominent pattern recognition techniques, Random Forests (RF) and Support Vector Machines (SVM) on the problem of identifying the stem and/or calyx region of Jonagold apples. This work demonstrates the inherit feature selection characteristics of the RF algorithm that drives its performance to rival a feature-tuned SVM. Although the SVM slightly outperforms the RF algorithm in classification status, more information about the data can be acquired through RF methodology.
Important Links:
Go Back
