ICDAR2024 MapText Competition (English)

We host ICDAR2024 MapText Competition, aiming to detect, recognize, and link text on diverse historical maps.

In the competition, we propose a new map benchmark dataset, containing 940 images crops (2K×2K pixel tiles), selected from ~900 distinct representative maps in the David Rumsey Collection. The David Rumsey Collection hosts an extensive set of over 126,000 maps spanning maps from the 16th to the 21st century that encompass regions from every continent, the Pacific, the Arctic, and the entirety of the World. See more details.

Train Data Link: https://s3.msi.umn.edu/icdar24-competition-data/icdar24-train-png.zip

Validation Data Link: https://s3.msi.umn.edu/icdar24-competition-data/icdar24-val-png.zip

Test Data Link: TBA

Citation

Li, Z., Lin, Y., Chiang, Y.Y., Weinman, J., Tual, S., Chazalon, J., Perret, J., Duménieu, B. and Abadie, N., 2024, August. ICDAR 2024 competition on historical map text detection, recognition, and linking. In International Conference on Document Analysis and Recognition (pp. 363-380). Cham: Springer Nature Switzerland.

Annotation Details

Please refer to the guideline document for annotation details. Contact Yijun Lin or Zekun Li if you have any questions.