Chinese Street View Text: Large-Scale Chinese Text Reading With Partially Supervised Learning

Sun, Yuliang; Liu, Jiaming; Liu, Wei; Han, Junyu; Ding, Errui; Liu, Jingtuo

doi:10.1109/iccv.2019.00918

Cited by 47 publications

(18 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We generate a synthetic long text dataset with the engine in [9], which includes 3 million images. Besides, we also use the training set of RCTW [32] and LSVT [33] as training data. Following the configuration described in Sec.…”

Section: Results On Non-latin Long Textmentioning

confidence: 99%

Towards Accurate Scene Text Recognition with Semantic Reasoning Networks

Yu¹,

Li²,

Zhang³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Scene text image contains two levels of contents: visual texture and semantic information. Although the previous scene text recognition methods have made great progress over the past few years, the research on mining semantic information to assist text recognition attracts less attention, only RNN-like structures are explored to implicitly model semantic information. However, we observe that RNN based methods have some obvious shortcomings, such as time-dependent decoding manner and one-way serial transmission of semantic context, which greatly limit the help of semantic information and the computation efficiency. To mitigate these limitations, we propose a novel end-to-end trainable framework named semantic reasoning network (SRN) for accurate scene text recognition, where a global semantic reasoning module (GSRM) is introduced to capture global semantic context through multi-way parallel transmission. The state-of-the-art results on 7 public benchmarks, including regular text, irregular text and non-Latin long text, verify the effectiveness and robustness of the proposed method. In addition, the speed of SRN has significant advantages over the RNN based methods, demonstrating its value in practical use.

show abstract

Section: Results On Non-latin Long Textmentioning

confidence: 99%

Towards Accurate Scene Text Recognition with Semantic Reasoning Networks

Yu¹,

Li²,

Zhang³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Second, we see potential in using a transformer in an end-to-end text spotting system. Third, TRIG can be improved to solve the problem about other languages [54] and hand-written (cursive) text recognition [55,56].…”

Section: Discussionmentioning

confidence: 99%

TRIG: Transformer-Based Text Recognizer with Initial Embedding Guidance

Yue

Jia

et al. 2021

Electronics

View full text Add to dashboard Cite

Scene text recognition (STR) is an important bridge between images and text, attracting abundant research attention. While convolutional neural networks (CNNS) have achieved remarkable progress in this task, most of the existing works need an extra module (context modeling module) to help CNN to capture global dependencies to solve the inductive bias and strengthen the relationship between text features. Recently, the transformer has been proposed as a promising network for global context modeling by self-attention mechanism, but one of the main short-comings, when applied to recognition, is the efficiency. We propose a 1-D split to address the challenges of complexity and replace the CNN with the transformer encoder to reduce the need for a context modeling module. Furthermore, recent methods use a frozen initial embedding to guide the decoder to decode the features to text, leading to a loss of accuracy. We propose to use a learnable initial embedding learned from the transformer encoder to make it adaptive to different input images. Above all, we introduce a novel architecture for text recognition, named TRansformer-based text recognizer with Initial embedding Guidance (TRIG), composed of three stages (transformation, feature extraction, and prediction). Extensive experiments show that our approach can achieve state-of-the-art on text recognition benchmarks.

show abstract

“…Most of these images were sourced from street boards and labeled as geometric shapes with varying vertices. A large-scale street view text (LSVT) of ICDAR 2019 [30] provides annotations to all 20,000 test images, 30,000 training images. ICDAR 2019 (MLT19) [31], this dataset comprises 18,000 images that have been annotated at the word level.…”

Section: Datasetsmentioning

confidence: 99%

ETDR: An Exploratory View of Text Detection and Recognition in Images and Videos

Lokkondra¹,

Ramegowda²,

Thimmaiah³

et al. 2021

RIA

View full text Add to dashboard Cite

Images and videos with text content are a direct source of information. Today, there is a high need for image and video data that can be intelligently analyzed. A growing number of researchers are focusing on text identification, making it a hot issue in machine vision research. Since this opens the way, several real-time-based applications such as text detection, localization, and tracking have become more prevalent in text analysis systems. To find out more about how text information may be extracted, have a look at our survey. This study presents a trustworthy dataset for text identification in images and videos at first. The second part of the article details the numerous text formats, both in images and video. Third, the process flow for extracting information from the text and the existing machine learning and deep learning techniques used to train the model was described. Fourth, explain assessment measures that are used to validate the model. Finally, it integrates the uses and difficulties of text extraction across a wide range of fields. Difficulties focus on the most frequent challenges faced in the actual world, such as capturing techniques, lightning, and environmental conditions. Images and videos have evolved into valuable sources of data. The text inside the images and video provides a massive quantity of facts and statistics. However, such data is not easy to access. This exploratory view provides easier and more accurate mathematical modeling and evaluation techniques to retrieve the text in image and video into an accessible form.

show abstract

Chinese Street View Text: Large-Scale Chinese Text Reading With Partially Supervised Learning

Cited by 47 publications

References 36 publications

Towards Accurate Scene Text Recognition with Semantic Reasoning Networks

Towards Accurate Scene Text Recognition with Semantic Reasoning Networks

TRIG: Transformer-Based Text Recognizer with Initial Embedding Guidance

ETDR: An Exploratory View of Text Detection and Recognition in Images and Videos

Contact Info

Product

Resources

About