Classification of 3D Archaeological Objects Using Multi-View Curvature Structure Signatures

Abstract

We propose a generalized 3D shape descriptor for the efficient classification of 3D archaeological artifacts. Our descriptor is based on a multi-view approach of curvature features, consisting of the following steps: pose normalization of 3D models, local curvature descriptor calculation, construction of 3D shape descriptor using the multi-view approach and curvature maps, and dimensionality reduction by random projections. We generate two descriptors from two different paradigms: 1) handcrafted, wherein the descriptor is manually designed for object feature extraction, and directly passed on to the classifier and 2) machine learnt, in which the descriptor automatically learns the object features through a pretrained deep neural network model (VGG-16) for transfer learning and passed on to the classifier. These descriptors are applied to two different archaeological datasets: 1) non-public Mexican dataset, represented by a collection of 963 3D archaeological objects from the Templo Mayor Museum in México City that includes anthropomorphic sculptures, figurines, masks, ceramic vessels, and musical instruments; and 2) 3D pottery content-based retrieval benchmark dataset, consisting of 411 objects. Once the multi-view descriptors are obtained, we evaluate their effectiveness by using the following object classification schemes: K -nearest neighbor, support vector machine, and structured support vector machine. Our object descriptors classification results are compared against five popular 3D descriptors in the literature, namely, rotation invariant spherical harmonic, histogram of spherical orientations, signature of histograms of orientations, symmetry descriptor, and reflective symmetry descriptor. Experimentally, we were able to verify that our machine learnt and handcrafted descriptors offer the best classification accuracy (20% better on average than comparative descriptors), independently of the classification methods. Our proposed descriptors are able to capture sufficient information to discern among different classes, concluding that it adequately characterizes the datasets.INDEX TERMS 3D shape descriptor, multi-class classification, multi-view approach, curvature, transfer learning.

ADMorph: A 3D Digital Microfossil Morphology Dataset for Deep Learning

“…By data preprocessing, including the PCA and CM calculations, the experimental result increased 0.82%. Evidence has shown that accuracy improves using a CM when classifying archeological data [43]. Toward the improvement of feature extraction, including the FT and LSTM-RNN models [60], the accuracy of 3D microfossil model classification increased by 5.56%.…”

“…Mean curvature is a single shape indicator that characterizes local variations of a surface. It is widely used in different applications, such as 3D face recognition [41], 3D skeleton shapes [42], and 3D object classification [43]. The mean curvature function is defined as follows:…”

A 3D digitisation workflow for architecture-specific annotation of built heritage

“…One of the most widely used applications of computational methods in human sciences is the automatic annotation and classification of large datasets (of text, images, etc. ) in digital libraries that otherwise would require a highly laborious annotation process by trained and skilled users (Canul-Ku et al, 2018;Engel et al, 2019;Dhali et al, 2020). In the case of image-based datasets, computer vision methods have been used to analyse and annotate photos e.g., the geo-reference of series of aerial photos (Cantoro, 2014), the semantic analysis of a digital library of museum artefacts, or a collection of old photographs for the semantic categorisation of contents (Eramian et al, 2017).…”

magazén

“…the classification of artefacts based on their shape. There is a large number of studies on this subject, those more relevant for archaeology are probably the ones by Tal (2014), Canul Ku et al (2018) Hermon et al (2018), and the research carried out in the EU-funded project GRAVITATE 11 coordinated in 2015-2018 by Michela Spagnolo of CNR (Consiglio Nazionale delle Ricerche) as well as in the AR-CHAIDE 12 EU-funded project.…”