Saved in:
Bibliographic Details
Main Authors: Su, Yuchen, Chen, Zhineng, Shao, Zhiwen, Du, Yuning, Ji, Zhilong, Bai, Jinfeng, Zhou, Yong, Jiang, Yu-Gang
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2306.15142
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913206332555264
author Su, Yuchen
Chen, Zhineng
Shao, Zhiwen
Du, Yuning
Ji, Zhilong
Bai, Jinfeng
Zhou, Yong
Jiang, Yu-Gang
author_facet Su, Yuchen
Chen, Zhineng
Shao, Zhiwen
Du, Yuning
Ji, Zhilong
Bai, Jinfeng
Zhou, Yong
Jiang, Yu-Gang
contents Recently, regression-based methods, which predict parameterized text shapes for text localization, have gained popularity in scene text detection. However, the existing parameterized text shape methods still have limitations in modeling arbitrary-shaped texts due to ignoring the utilization of text-specific shape information. Moreover, the time consumption of the entire pipeline has been largely overlooked, leading to a suboptimal overall inference speed. To address these issues, we first propose a novel parameterized text shape method based on low-rank approximation. Unlike other shape representation methods that employ data-irrelevant parameterization, our approach utilizes singular value decomposition and reconstructs the text shape using a few eigenvectors learned from labeled text contours. By exploring the shape correlation among different text contours, our method achieves consistency, compactness, simplicity, and robustness in shape representation. Next, we propose a dual assignment scheme for speed acceleration. It adopts a sparse assignment branch to accelerate the inference speed, and meanwhile, provides ample supervised signals for training through a dense assignment branch. Building upon these designs, we implement an accurate and efficient arbitrary-shaped text detector named LRANet. Extensive experiments are conducted on several challenging benchmarks, demonstrating the superior accuracy and efficiency of LRANet compared to state-of-the-art methods. Code is available at: \url{https://github.com/ychensu/LRANet.git}
format Preprint
id arxiv_https___arxiv_org_abs_2306_15142
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network
Su, Yuchen
Chen, Zhineng
Shao, Zhiwen
Du, Yuning
Ji, Zhilong
Bai, Jinfeng
Zhou, Yong
Jiang, Yu-Gang
Computer Vision and Pattern Recognition
Recently, regression-based methods, which predict parameterized text shapes for text localization, have gained popularity in scene text detection. However, the existing parameterized text shape methods still have limitations in modeling arbitrary-shaped texts due to ignoring the utilization of text-specific shape information. Moreover, the time consumption of the entire pipeline has been largely overlooked, leading to a suboptimal overall inference speed. To address these issues, we first propose a novel parameterized text shape method based on low-rank approximation. Unlike other shape representation methods that employ data-irrelevant parameterization, our approach utilizes singular value decomposition and reconstructs the text shape using a few eigenvectors learned from labeled text contours. By exploring the shape correlation among different text contours, our method achieves consistency, compactness, simplicity, and robustness in shape representation. Next, we propose a dual assignment scheme for speed acceleration. It adopts a sparse assignment branch to accelerate the inference speed, and meanwhile, provides ample supervised signals for training through a dense assignment branch. Building upon these designs, we implement an accurate and efficient arbitrary-shaped text detector named LRANet. Extensive experiments are conducted on several challenging benchmarks, demonstrating the superior accuracy and efficiency of LRANet compared to state-of-the-art methods. Code is available at: \url{https://github.com/ychensu/LRANet.git}
title LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2306.15142