HamdiJr/Egyptian_hieroglyphs

Dataset Overview

Dataset Name

• Egyptian hieroglyphs 𓂀
• Hieroglyphs image dataset along with Language Model

Dataset Features

• Source: Built from 10 images in the book The Pyramid of Unas (Alexandre Piankoff, 1955).
• Image Numbers: 3, 5, 7, 9, 20, 21, 22, 23, 39, 41.
• Annotation: Every hieroglyph was manually annotated and labeled according to the Gardiner Sign List.
• Image Naming: File names contain the label and image index.

Dataset Statistics

• Total Images: 4,210 (including 179 labeled as UNKNOWN).
• Total Classes: 171 (excluding the UNKNOWN class).

Annotation Accuracy

• Note: Annotations may not be fully accurate; unidentified hieroglyphs are marked as “UNKNOWN”.

Data Processing

• Manual Annotation: Hand‑annotated hieroglyphs.
• Automated Detection: Automatic extraction of hieroglyphs using text detection methods, stored in Dataset/Automated/.
• Location Information: x/y coordinates for each hieroglyph stored in the Location-folder.

Dataset Structure

• Images: The 10 source images from The Pyramid of Unas.
• Manual Annotation: Hieroglyph image crops with location data.
• Automated Detection: Automatically detected hieroglyph crops with location data.
• Language Model: Egyptian text, dictionary, and n‑grams from the JSesh database.

License

• GPL: Non‑commercial use only.