下载链接：

https://dataverse.harvard.edu/citation?persistentId=doi:10.7910/DVN/C308X3

下载链接

链接失效反馈

官方服务：

资源简介：

<!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <title>Ancient Egyptian Hieroglyphic Datasets</title> </head> <body> <section id="datasets"> <h2>Datasets for Ancient Egyptian Hieroglyphic Research</h2> <section id="hla-dataset"> <h3>Hieroglyphic Layout Analysis (HLA) Dataset</h3> Overview: The Hieroglyphic Layout Analysis (HLA) Dataset is a unique resource comprising 897 high-resolution images, each containing multiple lines of Ancient Egyptian hieroglyphs and cartouches. This dataset is specifically designed for tasks related to identifying and segmenting the layout of hieroglyphic texts within artifact images. Data Collection: The HLA Dataset was meticulously compiled from diverse sources: <ul> <li>Direct Collection (Egyptian Museum in Cairo, Egypt) : During on-site visits, we captured 230 high-resolution images of various artifacts, including Pharaonic coffins, statues, and wall inscriptions housed in the Egyptian Museum in Cairo.</li> <li>Online Museum Repositories: We leveraged the growing trend of museums providing open access to their collections, gathering over 300 images from The Metropolitan Museum of Art (“The Met”), New York, USA . The Met’s extensive collection spans approximately 30,000 objects of artistic, historical, and cultural significance, dating from around 300,000 BCE to the 4th century CE.</li> <li>Museo Egizio, Turin, Italy: We also collected more than 200 images from the Egyptian Museum in Turin (“Museo Egizio”), which has made over 4,000 of its approximately 40,000 objects accessible online.</li> <li>Specific Archaeological Sites: The dataset includes 18 images of the wall inscriptions within the Unas pyramid in Giza, Egypt, known for their dense concentration of hieroglyphic signs.</li> <li>Additional Online Resources: Further images were sourced from the online collections of: <ul> <li>The French Institute for Oriental Archaeology (IFAO) in Cairo</li> <li>The Rosicrucian Egyptian Museum (REM), USA</li> <li>The British Museum, UK</li> <li>The Museum of Fine Arts in Boston, USA</li> <li>The Louvre Museum in Paris</li> <li>The Egyptian Museum and Papyrus Collection of Berlin, Germany</li> <li>General online searches for relevant artifacts.</li> </ul> </li> </ul> Annotation Details: Following image collection and rigorous selection, every instance within the HLA Dataset underwent manual segmentation and annotation using the CVAT platform . Polygonal segmentation masks were created for two primary classes: “Line” (representing rows of hieroglyphs) and “Cartouche” (oval enclosures containing royal names). Individual images within the dataset can contain a significant number of these objects, with some images featuring up to 152 distinct lines or cartouches. Key Statistics: <ul> <li>Total Images: 897</li> <li>Annotation Classes: 2 (“Line”, “Cartouche”)</li> <li>Annotation Type: Polygon Segmentation Masks</li> <li>Maximum Objects per Image: 152</li> </ul> Potential Uses: This dataset is ideal for training and evaluating models for: <ul> <li>Hieroglyphic line detection and segmentation.</li> <li>Cartouche detection and segmentation.</li> <li>Layout analysis of hieroglyphic inscriptions on various artifact types.</li> </ul> Json annotation files "in coco format": <ul> <li>Train: 705 images.</li> <li>Validation: 178 images.</li> <li>Test: 10 images.</li> </ul> </section> </section> </body> </html>

应用场景：

Egyptian Hieroglyphic Layout Analysis