Transformer co-attention for word segmentation in image data

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23520%2F22%3A43965718" target="_blank" >RIV/49777513:23520/22:43965718 - isvavai.cz</a>

Result on the web

<a href="http://hdl.handle.net/11025/48796" target="_blank" >http://hdl.handle.net/11025/48796</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Transformer co-attention for word segmentation in image data

Original language description

Despite the huge progress in the past years, optical character recognition (OCR) of handwritten text is still a hard task. A prerequisite to it is extracting parts of the image that contain the text–line segmentation. In this work, wepresumethatlinesegmentationisavailable and are interested in splitting the lines into individual words – word segmentation. Such word segmentation can have many uses e.g. highlighting relevant words on the page when performing full text search or supplementing the full text search with search based on visual similarity between handwritten words.

Czech name

—

Czech description

—

Type

O - Miscellaneous

CEP classification

—

OECD FORD branch

20205 - Automation and control systems

Project

—

Continuities

S - Specificky vyzkum na vysokych skolach

Publication year

2022

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů