Abstract
Art valuation is left to the connoisseurs, collectors, and curators. Gaining expert knowledge takes years of mastery and professional opinions often come with high costs. As such, it is quite expensive for new artists and new art enthusiasts to estimate an initial value for the work or art collection.
Using about 3,000 images of Filipino artwork, this study provides a framework to estimate the price range of Filipino paintings. A seven-layer convolutional neural network is implemented and tuned to achieve a 62% F1-score accuracy in estimating the price range of a painting. This model serves as a price guide to artists and collectors in valuing Filipino paintings.
