Hieroglyphs AI allows you to automatically identify the Gardiner codes of hieroglyphs contained in image files on your computer using a deep convolutional neural network.

Click the screen, hold the left mouse button and move the mouse to draw a square enclosing the hieroglyph of interest. For successful recognition, an Egyptian sign should be approximately in the middle of the selected area and should be fully represented (without being cut off).

The neural network was trained on about 142,000 hand-labelled hieroglyphs of 706 classes from stelae, temple walls, tombs, papyri and sign lists. It shows the five most likely Gardiner codes with their probabilities.

On monuments it has never seen, the correct code is the first answer in about 88% of cases and among the top five in about 96%.

Questions, more recognitions for your research, or labelled data to share? Write to info@hieroglyphs.ai.