FC2RN: A Fully Convolutional Corner Refinement Network for Accurate Multi-Oriented Scene Text Detection

Qin, Xugong; Zhou, Yu; Guo, Youhui; Wu, Dayan; Wang, Weiping

doi:10.1109/icassp39728.2021.9413821

Cited by 20 publications

(5 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…TotalText is a dataset that includes horizontal, oriented, and [23] 76.8 73.1 75.0 86.0 70.0 77.0 ---RRPN [17] 82.0 73.0 77.0 82.0 68.0 74.0 ---TextBoxes++ [16] 87.8 78.5 82.9 ------TextSnake [13] 84.9 80.4 82.6 83.2 73.9 78.3 82.7 74.5 78.4 CRAFT [14] 89.9 84.3 86.9 88.2 78.2 82.9 87.6 79.9 83.6 PSENet + STKM [24] [11,26,27], we include extra 400 images from HUST-TR400 for training.…”

Section: Datasetsmentioning

confidence: 99%

UNITS: Unsupervised Intermediate Training Stage for Scene Text Detection

Guo¹,

Zhou²,

Qin³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Recent scene text detection methods are almost based on deep learning and data-driven. Synthetic data is commonly adopted for pre-training due to expensive annotation cost. However, there are obvious domain discrepancies between synthetic data and real-world data. It may lead to suboptimal performance to directly adopt the model initialized by synthetic data in the fine-tuning stage. In this paper, we propose a new training paradigm for scene text detection, which introduces an UNsupervised Intermediate Training Stage (UNITS) that builds a buffer path to real-world data and can alleviate the gap between the pre-training stage and finetuning stage. Three training strategies are further explored to perceive information from real-world data in an unsupervised way. With UNITS, scene text detectors are improved without introducing any parameters and computations during inference. Extensive experimental results show consistent performance improvements on three public datasets.

show abstract

Section: Datasetsmentioning

confidence: 99%

UNITS: Unsupervised Intermediate Training Stage for Scene Text Detection

Guo¹,

Zhou²,

Qin³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Through a complete analysis of fake news dataset prior to preparing this paper, we find some common defects about fake news, which can be categorized as presentation defects. Literally, "presentation defect" denotes the instantly visible defect in news article presentations, which widely exists in each presentation modality, like titles, textual contents [11][12] [13]. Specifically, "presentation defect" covers information consistency defects among these news.…”

Section: Introductionmentioning

confidence: 99%

APSN: Adversarial Pseudo-Siamese Network for Fake News Stance Detection

Zhou

Yang

Huang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Fake news is a longstanding issue that has existed on the social network, whose negative impact has been increasingly recognized since the 2016 US presidential election. During the election, numerous fake news about the candidates distributes vastly in the online social networks. Identifying inauthentic news quickly is an essential purpose for this research to enhance the trustworthiness of news in online social networks, which will be the task studied in this paper. The fake news stance detection can contribute to detect a startling amount of fake news, which aims at evaluating the relevance between the headline and text bodies. There exists a significant difference between news article headline and text body, since headlines with several key phrases are usually much shorter than the text bodies. Such an information imbalance challenge may cause serious problems for the stance detection task. What’s more, news article data in online social networks is usually exposed to various types of noise and can be contaminated, which poses more challenges for the stance detection task. In this paper, we propose a novel fake news stance detection model, namely Adversarial PseudoSiamese Network model (APSN), to solve these challenges. With coupled input components with imbalanced parameters, APSN can learn and compute feature vectors and similarity score of news article headlines and text bodies simultaneously. In addition, by adopting adversarial setting, besides the regular training set, a set of noisy training instances will be generated and fed to APSN in the learning process, which can significantly enhance the robustness of the model. Extensive experiments have been done on real-world fake news dataset, and the experimental results reveal that the presented model exceeds compared suspicious information detection models with significant advantages.

show abstract

“…Usually, text regions are In literature, scene text reading manuscripts can be split into two groups. The one way is using the deep learning method detecting text blocks instead of individual characters and then grouping them into text lines [2]- [11], [12]- [15]. The other way is the traditional method localizing individual characters first and then grouped into words.…”

Section: Introductionmentioning

confidence: 99%

Dominating set based arbitrary oriented bilingual scene text localization

Jayanth¹,

Kapanaiah²

2022

IJECE

View full text Add to dashboard Cite

Localizing and recognizing arbitrarily oriented text in natural scene images is the biggest challenge. It is because scene texts are often erratic in shapes. This paper presents a simple and effective graph representational algorithm for detecting arbitrary-oriented text location to smoothen the text recognition process because of its high impact and simplicity of representation. An arbitrarily oriented text can be horizontal, vertical, perspective, curved (diagonal/off-diagonal), or even a combination. As a pre-processing step, image enhancement is performed in the frequency domain to improve the representation of images that are invariant to intensity. It is necessary to draw bounding boxes for each candidate character in the scene images to extract text regions. This step is carried out by utilizing the advantage of the region-based approach called maximally stable extremal regions. A typical problem with curved text localization is that non-text objects may occur within localized text regions. Our method is the first in the literature that searches for dominating sets to solve this problem. This dominating set method outperforms several traditional methods, including deep learning methods used for arbitrary text localization, on challenging datasets like 13th international conference on document analysis and recognition (ICDAR 2015), multi-script robust reading competition (MRRC), CurvedText 80 (CUTE80), and arbitrary text (ArT).

show abstract

FC²RN: A Fully Convolutional Corner Refinement Network for Accurate Multi-Oriented Scene Text Detection

Cited by 20 publications

References 25 publications

UNITS: Unsupervised Intermediate Training Stage for Scene Text Detection

UNITS: Unsupervised Intermediate Training Stage for Scene Text Detection

APSN: Adversarial Pseudo-Siamese Network for Fake News Stance Detection

Dominating set based arbitrary oriented bilingual scene text localization

Contact Info

Product

Resources

About