These are pixel annotations of regions in the Classification of Medieval Latin 
Manuscripts (CLaMM) 2016 competition training set images.  They were made using
the PixLabeler tool and Otsu binarization.  The annotations are not perfect as
their intended use was to judge whether extracted patches were of mostly
textual content or not.

In pixel_labels.zip, there are grayscale images in PNG format with intesities 
0-15 corresponding to pixel classes.  In pixel_labels_color.zip, these pixel 
classes are assigned a visual color for easy viewing. These labels are:

0:(255,0,0) # caroline
1:(0,255,0) # cursiva
2:(0,0,128) # half uncial
3:(0,0,255) # humanistic
4:(128,128,0) # humanistic_cursiva
5:(128,0,128) # hybrida
6:(128,0,0) # praegothica
7:(255,128,128)  # semihybrida
8:(128,255,128)  # semitextualis
9:(0,128,0) # southern textualis
10:(128,128,255)  # textualis
11:(0,128,128) # uncial
12:(255,128,64)  # figure
13:(141,56,201)  # block letter
14:(37,65,23)  # side tag
15:(0,0,0) # background (unlabeled)
15:(255,255,255) # background (background)
15:(128,5,23) # background (noise)

Citations:
	Annotations


	CLaMM Competition
	Cloppet, Florence, et al. "ICFHR2016 Competition on the Classification of Medieval Handwritings in Latin Script." International Conference in Frontiers on Handwriting Recognition ICFHR 2016. 2016.

	Pix Labeler
	Saund, Eric, Jing Lin, and Prateek Sarkar. "Pixlabeler: User interface for pixel-level labeling of elements in document images." Document Analysis and Recognition, 2009. ICDAR'09. 10th International Conference on. IEEE, 2009.