Label Embedding: A Frugal Baseline for Text Recognition

Rodríguez-Serrano, José A.; Gordo, Albert; Perronnin, Florent

doi:10.1007/s11263-014-0793-6

Cited by 105 publications

(54 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, there has been a lot of work surrounding the connection between image representations and word embeddings, both with natural images [8], and word images [1], [2], [9], [10]. For the handwritten word spotting community, hand-crafted image representations, primarily Fisher vectors and other bag-of-visual-words models, have been the features of choice [1], [3], [10].…”

Section: Related Workmentioning

confidence: 99%

“…Previous approaches use techniques like Canonical Correlation Analysis [1], Latent Semantic Analysis [9], or structured Support Vector Machines [9] to learn a mapping from image representation to a word embedding space. In order to allow for fine-tuning of the triplet-CNN, we opt for a small fully connected neural network, consisting of two hidden layers of size 4096.…”

Section: Learning the Embeddingmentioning

confidence: 99%

See 1 more Smart Citation

Semantic and Verbatim Word Spotting Using Deep Neural Networks

Wilkinson

Brun

2016

2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)

View full text Add to dashboard Cite

Abstract-In the last few years, deep convolutional neural networks have become ubiquitous in computer vision, achieving state-of-the-art results on problems like object detection, semantic segmentation, and image captioning. However, theyhave not yet been widely investigated in the document analysis community. In this paper, we present a word spotting system based on convolutional neural networks. We train a network to extract a powerful image representation, which we then embed into a word embedding space. This allows us to perform word spotting using both query-by-string and query-by-example in a variety of word embedding spaces, both learned and handcrafted, for verbatim as well as semantic word spotting. Our novel approach is versatile and the evaluation shows that it outperforms the previous state-of-the-art for word spotting on standard datasets.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Learning the Embeddingmentioning

confidence: 99%

Semantic and Verbatim Word Spotting Using Deep Neural Networks

Wilkinson

Brun

2016

2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)

View full text Add to dashboard Cite

show abstract

“…1 Traditionally, word spotting and recognition have focused on document images [6], [7], [8], [9], [10], [11], [12], [13], [14], where the main challenges come from differences in writing styles: the writing styles of different writers may be completely different for the same word. Recently, however, with the development of powerful computer vision techniques during the last decade, there has been an increased interest in performing word spotting and recognition on natural images [3], [4], [5], [15], [16], [17], which poses different challenges such as huge variations in illumination, point of view, typography, etc.…”

Section: Introductionmentioning

confidence: 99%

Word Spotting and Recognition with Embedded Attributes

Almazán

Gordo

Fornés

et al. 2014

IEEE Trans. Pattern Anal. Mach. Intell.

Self Cite

484

420

View full text Add to dashboard Cite

This paper addresses the problems of word spotting and word recognition on images. In word spotting, the goal is to find all instances of a query word in a dataset of images. In recognition, the goal is to recognize the content of the word image, usually aided by a dictionary or lexicon. We describe an approach in which both word images and text strings are embedded in a common vectorial subspace. This is achieved by a combination of label embedding and attributes learning, and a common subspace regression. In this subspace, images and strings that represent the same word are close together, allowing one to cast recognition and retrieval tasks as a nearest neighbor problem. Contrary to most other existing methods, our representation has a fixed length, is low dimensional, and is very fast to compute and, especially, to compare. We test our approach on four public datasets of both handwritten documents and natural images showing results comparable or better than the state-of-the-art on spotting and recognition tasks.

show abstract

“…Furthermore, after CCA the attribute scores and binary attributes lie in a more correlated space, which makes the comparison between the scores and the PHOCs for our QBS problem more principled. CCA can also be seen as a label embedding method, similar in spirit to the recent approach of [27]. CCA is also used as a dimensionality reduction tool: we reduce the dimensionality from 384 down to 192-256 dimensions.…”

Section: Calibration Of Scoresmentioning

confidence: 99%

“…A naive implementation of this attributes representation greatly outperforms the direct use of FVs. A very similar string embedding has been simultaneously proposed in [27]. However, in their case, the representation is used in a label embedding context, and not as a source of attributes.…”

Section: Introductionmentioning

confidence: 99%

Handwritten Word Spotting with Corrected Attributes

Almazán

Gordo

Fornés

et al. 2013

2013 IEEE International Conference on Computer Vision

Self Cite

View full text Add to dashboard Cite

We propose an approach to multi-writer word spotting, where the goal is to find a query word in a dataset comprised of document images. We propose an attributes-based approach that leads to a low-dimensional, fixed-length representation of the word images that is fast to compute and, especially, fast to compare. This approach naturally leads to an unified representation of word images and strings, which seamlessly allows one to indistinctly perform queryby-example, where the query is an image, and query-bystring, where the query is a string. We also propose a calibration scheme to correct the attributes scores based on Canonical Correlation Analysis that greatly improves the results on a challenging dataset. We test our approach on two public datasets showing state-of-the-art results.

show abstract

Label Embedding: A Frugal Baseline for Text Recognition

Cited by 105 publications

References 43 publications

Semantic and Verbatim Word Spotting Using Deep Neural Networks

Semantic and Verbatim Word Spotting Using Deep Neural Networks

Word Spotting and Recognition with Embedded Attributes

Handwritten Word Spotting with Corrected Attributes

Contact Info

Product

Resources

About