Conference item icon

Conference item

Deep structured output learning for unconstrained text recognition

Abstract:
We develop a representation suitable for the unconstrained recognition of words in natural images: the general case of no fixed lexicon and unknown length. To this end we propose a convolutional neural network (CNN) based architecture which incorporates a Conditional Random Field (CRF) graphical model, taking the whole word image as a single input. The unaries of the CRF are provided by a CNN that predicts characters at each position of the output, while higher order terms are provided by another CNN that detects the presence of N-grams. We show that this entire model (CRF, character predictor, N-gram predictor) can be jointly optimised by back-propagating the structured output loss, essentially requiring the system to perform multi-task learning, and training uses purely synthetically generated data. The resulting model is a more accurate system on standard real-world text recognition benchmarks than character prediction alone, setting a benchmark for systems that have not been trained on a particular lexicon. In addition, our model achieves state-of-the-art accuracy in lexicon-constrained scenarios, without being specifically modelled for constrained recognition. To test the generalisation of our model, we also perform experiments with random alpha-numeric strings to evaluate the method when no visual language model is applicable.
Publication status:
Not published
Peer review status:
Peer reviewed

Actions

Access Document

Files:
Publisher copy:
10.48550/arXiv.1412.5903

Authors

More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Role:
Author
ORCID:
0000-0002-8945-8573


Publisher:
International Conference on Learning Representations
Pages:
1-10
Publication date:
2015-05-09
Acceptance date:
2014-12-19
Event title:
3rd International Conference on Learning Representations (ICLR 2015)
Event location:
San Diego, CA, USA
Event website:
https://iclr.cc/archive/www/doku.php%3Fid=iclr2015:main.html
Event start date:
2015-05-07
Event end date:
2015-05-09
DOI:


Language:
English
Keywords:
Pubs id:
pubs:678958
UUID:
uuid:542ce865-628c-43b5-b308-595bcb5b7a06
Local pid:
pubs:678958
Source identifiers:
678958
Deposit date:
2017-02-09
ARK identifier:

Terms of use


Views and Downloads

Views and downloads will return soon






If you are the owner of this record, you can report an update to it here: Report update to this record

TO TOP