Translating handwritten bushman texts

Abstract

The Bleek and Lloyd Collection is a collection of artefacts documenting the life and language of the Bushman people of southern Africa in the 19th century. Included in this collection is a handwritten dictionary that contains English words and their corresponding |xam Bushman language translations. This dictionary allows for the manual translation of |xam words that appear in the notebooks of the Bleek and Lloyd collection. This, however, is not practical due to the size of the dictionary, which contains over 14000 entries. To solve this problem a content-based image retrieval system was built that allows for the selection of a |xam word from a notebook and returns matching words from the dictionary. The system shows promise with some search keys returning relevant results.

Creating a Handwriting Recognition Corpus for Bushman Languages

“…Fortunately though, some of it was recorded and written down by linguists Wilhelm Bleek, Lucy Lloyd and others [1] in the 19th century and this historical collection of notebooks, dictionaries and pieces of art have collectively come to be known as the Bleek and Lloyd Collection [1]. Systems have and are being built for preserving the notebooks [1] and dictionaries [10] as well as for enhancing interaction with the collection [11]. An example of part of a notebook page is shown in Figure 1.…”

A comparison of machine learning techniques for handwritten |Xam word recognition

“…The collection has been digitised and digital library systems have been created for preserving this collection and providing access to it via the Web [22,26]. Furthermore, some tools have been designed for enhanced interaction with the collection [27]. However, these systems are limited in that they mainly operate at the image level and not on the text contained in the notebooks.…”

Objetos complexos em bibliotecas digitais

“…A segunda abordagem, chamada Recuperac ¸ão por Conteúdo (Content-Based Image Retrieval -CBIR) [20] propõe a recuperac ¸ão automática de imagens, com base nas propriedades visuais, tais como textura, forma ou cor. Nesse sentido, imagens são usadas para procurar outras imagens em sistemas de CBIR [19,129]. Esta soluc ¸ão exige a construc ¸ão de descritores de imagens, que são caracterizadas por (i) um algoritmo de extrac ¸ão (descritor) para codificar características da imagem em vetores de características, e (ii) uma medida de similaridade para comparar duas imagens com base na distância entre os vetores de características.…”

“…Alguns tornaram-se produtos comerciais, outros foram propostos como protótipos de investigac ¸ão [60], sendo desenvolvidos em universidades e laboratórios de pesquisa. Existem ainda outros projetos que comparam descritores CBIR para melhorar o desempenho [98], para visualizar os resultados [126], combinar o conteúdo visual de imagens com metadados [21] ou mesmo explorar diferentes domínios [26,80,129]. Nenhuma dessas iniciativas porém, têm-se centrado em mecanismos para gerenciar objetos complexos de imagem.…”