TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text

Stroke order and velocity are helpful features in the fields of signature verification, handwriting recognition, and handwriting synthesis. Recovering these features from offline handwritten text is a challenging and well-studied problem. We propose a new model called TRACE (Trajectory Recovery by a...

Full description

Bibliographic Details
Main Author: Archibald, Taylor Neil
Format: Others
Published: BYU ScholarsArchive 2020
Subjects:
Online Access:https://scholarsarchive.byu.edu/etd/9299
https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=10308&context=etd
id ndltd-BGMYU2-oai-scholarsarchive.byu.edu-etd-10308
record_format oai_dc
spelling ndltd-BGMYU2-oai-scholarsarchive.byu.edu-etd-103082021-12-14T05:00:46Z TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text Archibald, Taylor Neil Stroke order and velocity are helpful features in the fields of signature verification, handwriting recognition, and handwriting synthesis. Recovering these features from offline handwritten text is a challenging and well-studied problem. We propose a new model called TRACE (Trajectory Recovery by an Adaptively-trained Convolutional Encoder). TRACE is a differentiable approach using a convolutional recurrent neural network (CRNN) to infer temporal stroke information from long lines of offline handwritten text with many characters. TRACE is perhaps the first system to be trained end-to-end on entire lines of text of arbitrary width and does not require the use of dynamic exemplars. Moreover, the system does not require images to undergo any pre-processing, nor do the predictions require any post-processing. Consequently, the recovered trajectory is differentiable and can be used as a loss function for other tasks, including synthesizing offline handwritten text. We demonstrate that temporal stroke information recovered by TRACE from offline data can be used for handwriting synthesis and establish the first benchmarks for a stroke trajectory recovery system trained on the IAM online database. 2020-12-01T08:00:00Z text application/pdf https://scholarsarchive.byu.edu/etd/9299 https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=10308&context=etd https://lib.byu.edu/about/copyright/ Theses and Dissertations BYU ScholarsArchive handwriting stroke trajectory recovery deep learning Computer Sciences
collection NDLTD
format Others
sources NDLTD
topic handwriting
stroke trajectory recovery
deep learning
Computer Sciences
spellingShingle handwriting
stroke trajectory recovery
deep learning
Computer Sciences
Archibald, Taylor Neil
TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
description Stroke order and velocity are helpful features in the fields of signature verification, handwriting recognition, and handwriting synthesis. Recovering these features from offline handwritten text is a challenging and well-studied problem. We propose a new model called TRACE (Trajectory Recovery by an Adaptively-trained Convolutional Encoder). TRACE is a differentiable approach using a convolutional recurrent neural network (CRNN) to infer temporal stroke information from long lines of offline handwritten text with many characters. TRACE is perhaps the first system to be trained end-to-end on entire lines of text of arbitrary width and does not require the use of dynamic exemplars. Moreover, the system does not require images to undergo any pre-processing, nor do the predictions require any post-processing. Consequently, the recovered trajectory is differentiable and can be used as a loss function for other tasks, including synthesizing offline handwritten text. We demonstrate that temporal stroke information recovered by TRACE from offline data can be used for handwriting synthesis and establish the first benchmarks for a stroke trajectory recovery system trained on the IAM online database.
author Archibald, Taylor Neil
author_facet Archibald, Taylor Neil
author_sort Archibald, Taylor Neil
title TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
title_short TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
title_full TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
title_fullStr TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
title_full_unstemmed TRACE: A Differentiable Approach to Line-Level Stroke Recovery for Offline Handwritten Text
title_sort trace: a differentiable approach to line-level stroke recovery for offline handwritten text
publisher BYU ScholarsArchive
publishDate 2020
url https://scholarsarchive.byu.edu/etd/9299
https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=10308&context=etd
work_keys_str_mv AT archibaldtaylorneil traceadifferentiableapproachtolinelevelstrokerecoveryforofflinehandwrittentext
_version_ 1723964204295127040