ICDAR 2017

ICDAR2017 Competition on the Classification of Medieval Handwritings in Latin Script

UPDATE (1/03/2017)

The Training Data Set for the ICDAR2017 CLAMM competition is available. Please register (how to is here: Registration)

The proposed competition is the second round of the Competition on the Classification of Medieval Handwritings in Latin Script. The first round was organized at ICFHR 2016 [1]. It was followed by a reflexive seminar gathering researchers in Humanities and Computer Science to analyze the result[2].

Tasks under evaluation

The tasks to be evaluated in the present competition is the classification of 3000 images of Latin Scripts, from handwritten books dated 500 C.E. to 1600 C.E. A similar competition was organized for ICFHR2016 and provides the training data-set and a preliminary benchmark.
The organizers provide a data-set consisting of 5000 images with annotation of script types. They will evaluate the results on a 3000 additional images which are not included in the training data-set.
We propose four possible tasks:
– Task 1: Style classification
– Task 2: Style classification on heterogeneously encoded data
– Task 3: Manuscript dating
– Task 4: Manuscript dating on heterogeneously encoded data
All tasks are divided in two steps:
1) The participants will have to provide a normalized “distance matrix” between pairs of images
2) The participants will have to provide a normalized multi-weighted labeling to each image, that is a “belonging matrix”describing the degree to which each image is associated with a particular label
Participants are expected to provide the CSV-files according to the format that is described in section 2. There are 12 pre-defined classes according to the script style and 15 pre-defined classes according to the date.

More info:

Data set

Schedule

Outputs and Evaluation

Registration and access to data

References