Detección jerárquica de jeroglíficos egipcios frecuentes mediante YOLO11 y RF-DETR

The automatic interpretation of hieroglyphs on high-density stelae is a complex problem due to multiple factors such as linguistic variability, data scarcity, and the natural wear of the surfaces. To overcome the limitations of traditional methods and minimize the costly manual labeling of each symbol, a hierarchical detection strategy has been designed. This strategy involves first analyzing and locating the most frequent signs that provide the greatest coverage on the stela, such as birds and clusters of individual structural glyphs. These are then used as anchor points for a model capable of leveraging spatial context to place unknown hieroglyphs under a generic label. Thus, by combining two cutting-edge architectures, YOLO11 and RF-DETR, a unified strategy is proposed for isolating and labelling signs, enabling their recognition and translation by expert classification networks.