Investigation into color designs of product packaging through visual evaluations using machine learning methods

For a commodity, in addition to its quality, external package is also very essential. This paper briefly introduced the intelligent support vector machine (SVM) algorithm for color design of packaging. The features were extracted from photos packages using scale-invariant feature transform (SIFT), and was trained tested packaging ceramic products collected at crafts market as sample set. Two schemes designed this study used further test. SVM compared with back-propagation (BP) convolutional neural network (CNN) algorithm. results showed that three algorithms could evaluate packages, but more accurate than BP CNN evaluating imagery design, both samples craft scheme paper.