EE Seminar: Identification of similarities in archaeological collections using deep learning algorithms: a Levantine case study.

05 בפברואר 2020, 15:00 
Room 011, Kitot Building 

Speaker: Avi Resler

M.Sc. student under the supervision of Dr. Raja Giryes and Dr. Filipe Natalio

 

Wednesday, February 5th, 2020, at 15:00

Room 011, Kitot Bldg., Faculty of Engineering

 

Identification of similarities in archaeological collections using deep learning algorithms:  a Levantine case study.

 

Abstract

 Artefacts that are found in archaeological excavations are often recognized by experts, who compare their appearance to other labeled objects that they have seen before or present in archaeological catalogs. Since this procedure may be subjective, scientific methods that aid archaeologists have became increasingly popular.

We have developed two machine learning tools which capture the similarity between two artefacts or similarities between groups of artefacts based on their RGB images. For the first antique recognition tool, we used face recognition deep neural network architecture, to measure the "archaeological" distance between images. In the second community detection tool, we aggregate similarities between images and measure the distance between assemblages - i.e., group of images. Based on that we applied a network-theory community detection algorithm, to find groups of archaeological sites that are linked to each other.

To test our methods, we used a highly diverse dataset of Israeli antiques. This dataset is a good case study due to geographical proximity between archaeological sites and the presence of artefacts from a wide range of archaeological ages.

אוניברסיטת תל אביב עושה כל מאמץ לכבד זכויות יוצרים. אם בבעלותך זכויות יוצרים בתכנים שנמצאים פה ו/או השימוש
שנעשה בתכנים אלה לדעתך מפר זכויות, נא לפנות בהקדם לכתובת שכאן >>