Cuneiform Annotator
Gina Stavropoulou
KU Leuven, ESAT - VISICS
CASA ‘16, Geneva
One registration process, many visualization possibilities
Data & Research
Digitization of Cuneiform Tablets with the Minidomε
Focus: Classification &
Retrieval on Cuneiform Tablets.
Methods: Supervised &
Unsupervised.
Problems: Lack of (Training) Data & Segmentation Issues
Data & Research
Challenge: How to browse through the collections
based on the tablet’s content?
Text Based Approach
Input: text query (BA)
Content Based Image Retrieval
(CBIR)
Input: Image Query
Transliteration of NP2, NP3, NP5...
Output
Image Retrieval
Text Based Approach
Input: text query (BA)
Content Based Image Retrieval
(CBIR)
Input: Image Query
Transliteration of NP2, NP3, NP5...
Output
Image Retrieval
Text Based Approach
Input: text query (BA)
Content Based Image Retrieval
(CBIR)
Input: Image Query
Transliteration of NP2, NP3, NP5...
Output
Image Retrieval
Training Data
Projection Profile
Line Segmentation:
A* search
Problems: Overlapping signs, cracks, inconsistent dimensions of signs .
Segmentation
Sign Segmentation:
Drop fall OCR
methods
Both Segmentation and Annotation are necessary
Web based annotator to segment and annotate signs
Annotator