The site of ‘Grotta Poesia’ is one of the richest repositories of filiform post-Paleolithic rock art in southern Italy, as well as the largest archive of Messapian inscriptions known. In 2022, research activities in the field were resumed thanks to the «DigiRock» project, of which objectives, methods and preliminary results are presented in this report.The project’s overarching aim is to develop an artificial intelligence algorithm for recognizing and classifying the engravings found in ‘Grotta Poesia’. The first goal of the project was a 3D scan of the cave using a Terrestrial Laser Scanner (TLS) so as to reconstruct the cavity's floor plan, cross sections and obtain the basic topography, but also to test the accuracy of the TLS in the three-dimensional acquisition of engravings.In order to study the signs engraved on the cave walls and to train an AI algorithm capable of recognizing and classifying them, surveys for digital image acquisition were planned. Using Structure from Motion (SfM) and Multi-view Stereo (MVS) technology, it was conducted 3D photogrammetry on the replicas of the cave walls and, more specifically, on the individual engraved signs.Furthermore, another goal of the project was to create a searchable database of Italian rock art spanning from the Upper Paleolithic to the Bronze Age; currently, the ‘reference collection’ includes more than 1,000 figures (painted or engraved) from about 40 different sites, including shelters, caves and mobile artefacts. At present a significant portion of the rock art signs documented at ‘Grotta Poesia’ appears attributable to a period between the Neolithic and the Bronze Age. The conclusions derived from the study and analysis of the current dataset confirm the effectiveness of deep learning techniques, in particular the use of convolutional neural networks (CNN), in the study of filiform rock art from ‘Grotta Poesia’ cave walls.
Il sito della ‘Grotta Poesia’ è uno dei più ricchi contesti di arte rupestre filiforme post-paleolitica nel sud Italia, nonché il più grande archivio di iscrizioni messapiche conosciute. Nel 2022, le attività di ricerca sul campo sono riprese grazie al progetto «DigiRock», i cui obiettivi, metodi e risultati preliminari sono presentati in questo articolo. L’obiettivo generale del progetto è sviluppare un algoritmo di intelligenza artificiale per il riconoscimento e la classificazione delle incisioni presenti a ‘Grotta Poesia’. Il primo obiettivo del progetto è stato la scansione 3D della grotta utilizzando un Terrestrial Laser Scanner (TLS) per ricostruire la pianta della cavità, le sezioni trasversali e ottenere la topografia di base, ma anche per testare l’accuratezza del TLS nell'acquisizione tridimensionale delle incisioni. Per studiare i segni incisi sulle pareti della grotta e addestrare un algoritmo di intelligenza artificiale capace di riconoscerli e classificarli, sono stati pianificati rilievi per l’acquisizione di immagini digitali. Utilizzando la tecnologia Structure from Motion (SfM) e Multiview Stereo (MVS), è stata condotta la fotogrammetria 3D sulle repliche delle pareti della grotta e, più specificamente, sui singoli segni incisi. Inoltre, un altro obiettivo del progetto è stato creare un database consultabile dell’arte rupestre italiana che spazi dal Paleolitico Superiore all’età del Bronzo; attualmente, la ‘collezione di riferimento’ include più di 1.000 figure (dipinte o incise) provenienti da circa 40 siti diversi, tra cui ripari, grotte e manufatti mobili. Al momento, una parte significativa dei segni di arte rupestre documentati alla ‘Grotta Poesia’ sembra attribuibile a un periodo tra il Neolitico e l’età del Bronzo. Le conclusioni derivate dallo studio e dall’analisi del 212 dataset attuale confermano l’efficacia delle tecniche di deep learning, in particolare l’uso di reti neurali convoluzionali (CNN), nello studio dell'arte rupestre filiforme delle pareti di ‘Grotta Poesia’.
Scarano, T., Lago, G., Spada, I., Luparelli, A. (2026). DigiRock: Recognizing, Digitizing, and Analysing the Rock Art of ‘Grotta Poesia’ with Artificial Intelligence. Bicester : Archaeopress Publishing Ltd.
DigiRock: Recognizing, Digitizing, and Analysing the Rock Art of ‘Grotta Poesia’ with Artificial Intelligence
Giancarlo Lago;
2026
Abstract
The site of ‘Grotta Poesia’ is one of the richest repositories of filiform post-Paleolithic rock art in southern Italy, as well as the largest archive of Messapian inscriptions known. In 2022, research activities in the field were resumed thanks to the «DigiRock» project, of which objectives, methods and preliminary results are presented in this report.The project’s overarching aim is to develop an artificial intelligence algorithm for recognizing and classifying the engravings found in ‘Grotta Poesia’. The first goal of the project was a 3D scan of the cave using a Terrestrial Laser Scanner (TLS) so as to reconstruct the cavity's floor plan, cross sections and obtain the basic topography, but also to test the accuracy of the TLS in the three-dimensional acquisition of engravings.In order to study the signs engraved on the cave walls and to train an AI algorithm capable of recognizing and classifying them, surveys for digital image acquisition were planned. Using Structure from Motion (SfM) and Multi-view Stereo (MVS) technology, it was conducted 3D photogrammetry on the replicas of the cave walls and, more specifically, on the individual engraved signs.Furthermore, another goal of the project was to create a searchable database of Italian rock art spanning from the Upper Paleolithic to the Bronze Age; currently, the ‘reference collection’ includes more than 1,000 figures (painted or engraved) from about 40 different sites, including shelters, caves and mobile artefacts. At present a significant portion of the rock art signs documented at ‘Grotta Poesia’ appears attributable to a period between the Neolithic and the Bronze Age. The conclusions derived from the study and analysis of the current dataset confirm the effectiveness of deep learning techniques, in particular the use of convolutional neural networks (CNN), in the study of filiform rock art from ‘Grotta Poesia’ cave walls.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.



