A Dataset and a Novel Neural Approach for Optical Gregg Shorthand Recognition

Zhai, Fangzhou; Fan, Yue; Verma, Tejaswani; Sinha, Rupali; Klakow, Dietrich

doi:10.1007/978-3-030-00794-2_24

Cited by 3 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zhai et al [32] apply similar transforms to all of the above but do not sample the respective parameters. Instead, they use fixed, pre-defined values and generate a total of eight images (with fixed choice of augmentation parameters) for every input.…”

Section: Data Augmentation Methods For Handwritten Text Recognitionmentioning

confidence: 99%

“…Zhai et al [32] propose to use a CNN-based feature extractor, followed by a recurrent neural network to generate transliterations of individual words, written in Gregg's stenography.…”

Section: Handwritten Stenography Recognitionmentioning

confidence: 99%

“…Three models are trained with random rotations in the ranges [−1.5, 1.5] ("rot1.5") [24], [−5, 5] ("rot5") [30] and [−10, 10] ("rot10"). Following [32], one model each is trained with a fixed rotation of +2, respectively -2, degrees ("rot+2", "rot-2"). Lastly, one model each is trained with only positive, respectively negative rotations, in the ranges [0, 1.5] ("positive") and [−1.5, 0] ("negative").…”

Section: Examined Augmentationsmentioning

confidence: 99%

See 2 more Smart Citations

A Study of Augmentation Methods for Handwritten Stenography Recognition

Heil¹,

Breznik²

2023

Preprint

View full text Add to dashboard Cite

One of the factors limiting the performance of handwritten text recognition (HTR) for stenography is the small amount of annotated training data. To alleviate the problem of data scarcity, modern HTR methods often employ data augmentation. However, due to specifics of the stenographic script, such settings may not be directly applicable for stenography recognition. In this work, we study 22 classical augmentation techniques, most of which are commonly used for HTR of other scripts, such as Latin handwriting. Through extensive experiments, we identify a group of augmentations, including for example contained ranges of random rotation, shifts and scaling, that are beneficial to the use case of stenography recognition. Furthermore, a number of augmentation approaches, leading to a decrease in recognition performance, are identified. Our results are supported by statistical hypothesis testing. Links to the publicly available dataset and codebase are provided.

show abstract

Section: Data Augmentation Methods For Handwritten Text Recognitionmentioning

confidence: 99%

“…Zhai et al [32] propose to use a CNN-based feature extractor, followed by a recurrent neural network to generate transliterations of individual words, written in Gregg's stenography.…”

Section: Handwritten Stenography Recognitionmentioning

confidence: 99%

Section: Examined Augmentationsmentioning

confidence: 99%

See 1 more Smart Citation

A Study of Augmentation Methods for Handwritten Stenography Recognition

Heil¹,

Breznik²

2023

Preprint

View full text Add to dashboard Cite

show abstract

A Study of Augmentation Methods for Handwritten Stenography Recognition

Heil

Breznik

2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Handwritten stenography recognition and the LION dataset

Heil,

Nauwerck

2024

IJDAR

View full text Add to dashboard Cite

In this paper, we establish the first baseline for handwritten stenography recognition, using the novel LION dataset, and investigate the impact of including selected aspects of stenographic theory into the recognition process. We make the LION dataset publicly available with the aim of encouraging future research in handwritten stenography recognition. A state-of-the-art text recognition model is trained to establish a baseline. Stenographic domain knowledge is integrated by transforming the target sequences into representations which approximate diplomatic transcriptions, wherein each symbol in the script is represented by its own character in the transliteration, as opposed to corresponding combinations of characters from the Swedish alphabet. Four such encoding schemes are evaluated and results are further improved by integrating a pre-training scheme, based on synthetic data. The baseline model achieves an average test character error rate (CER) of 29.81% and a word error rate (WER) of 55.14%. Test error rates are reduced significantly (p< 0.01) by combining stenography-specific target sequence encodings with pre-training and fine-tuning, yielding CERs in the range of 24.5–26% and WERs of 44.8–48.2%. An analysis of selected recognition errors illustrates the challenges that the stenographic writing system poses to text recognition. This work establishes the first baseline for handwritten stenography recognition. Our proposed combination of integrating stenography-specific knowledge, in conjunction with pre-training and fine-tuning on synthetic data, yields considerable improvements. Together with our precursor study on the subject, this is the first work to apply modern handwritten text recognition to stenography. The dataset and our code are publicly available via Zenodo.

show abstract

A Dataset and a Novel Neural Approach for Optical Gregg Shorthand Recognition

Cited by 3 publications

References 9 publications

A Study of Augmentation Methods for Handwritten Stenography Recognition

A Study of Augmentation Methods for Handwritten Stenography Recognition

A Study of Augmentation Methods for Handwritten Stenography Recognition

Handwritten stenography recognition and the LION dataset

Contact Info

Product

Resources

About