Deepening in the 3D Modelling: Multisource Analysis of a Polychrome Ceramic Vessel Through the Integration of Thermal and Hyperspectral Information

Classification of engraved pottery sherds mixing deep-learning features by compact bilinear pooling

“…The new low-cost scanning technologies by laser scanner or photogrammetry promote the use of pattern recognition and artificial intelligence in particular to automate the heavy tasks of inventorying and classifying artifacts in the fields of cultural heritage (Gomes and al., 2014). For archaeological pottery fragments, studies have proposed a classification based either on the shape and profile (Willis et al, 2003;Kampel and Sablatnig, 2007;Son et al, 2013;Zheng et al, 2014;Zvietcovich et al, 2016), on the color and material characterization (Kampel and Sablatnig, 2000;Stanco and Gueli, 2013;Farjas et al, 2012;Smith et al, 2010;Abadi et al, 2012;Makridis and Daras, 2012) or on their decorations (Guarnera et al, 2011;Zhou et al, 2017Zhou et al, , 2018. Shape analysis is generally based on an ideal model of surface of revolution to extract an axis and a profile, then used to match the sherds, sometimes going as far as an automatic assembly of a pot from broken sherds (Willis et al, 2003).…”