Evaluation of Attention Mechanisms in U-Net Architectures for Semantic Segmentation of Brazilian Rock Art Petroglyphs

TL;DR

This paper compares three U-Net-based architectures with attention mechanisms for semantic segmentation of Brazilian rock art petroglyphs, demonstrating improved accuracy and recall over baseline models in archaeological image analysis.

Contribution

It introduces and evaluates three novel U-Net architectures incorporating attention mechanisms and a specialized loss function for archaeological petroglyph segmentation.

Findings

Attention-Residual BEGL-UNet achieved the highest Dice Score of 0.710.

Spatial Channel Attention BEGL-UNet achieved a Dice Score of 0.707.

All attention-based models outperformed the baseline by 2.5-2.9% in Dice Score.

Abstract

This study presents a comparative analysis of three U-Net-based architectures for semantic segmentation of rock art petroglyphs from Brazilian archaeological sites. The investigated architectures were: (1) BEGL-UNet with Border-Enhanced Gaussian Loss function; (2) Attention-Residual BEGL-UNet, incorporating residual blocks and gated attention mechanisms; and (3) Spatial Channel Attention BEGL-UNet, which employs spatial-channel attention modules based on Convolutional Block Attention Module. All implementations employed the BEGL loss function combining binary cross-entropy with Gaussian edge enhancement. Experiments were conducted on images from the Po\c{c}o da Bebidinha Archaeological Complex, Piau\'i, Brazil, using 5-fold cross-validation. Among the architectures, Attention-Residual BEGL-UNet achieved the best overall performance with Dice Score of 0.710, validation loss of 0.067, and highest recall of 0.854. Spatial Channel Attention BEGL-UNet obtained comparable performance with DSC of 0.707 and recall of 0.857. The baseline BEGL-UNet registered DSC of 0.690. These results demonstrate the effectiveness of attention mechanisms for archaeological heritage digital preservation, with Dice Score improvements of 2.5-2.9% over the baseline.