Curved Text Detection in Natural Scene Images with Semi- and Weakly-Supervised Learning

Dongbao Yang; Weiping Wang; Xugong Qin; Yu Zhou

REVIEW 3 major objections 6 minor 33 references

Curved Text Detection in Natural Scene Images with Semi- and Weakly-Supervised Learning

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a curved text detector can be trained to near state-of-the-art accuracy with only 10% pixel-level annotated images and 90% rectangle-annotated images, by generating pseudo masks from ground-truth boxes.

desk verdict A sensible pseudo-mask strategy for label-efficient curved text detection, but the headline claim is built on best-round test selection and a single split; worth reviewing with a request for a fixed-protocol rerun. read the letter →

arxiv 1908.09990 v1 pith:RADJRXBH submitted 2019-08-27 cs.CV

Xugong Qin , Yu Zhou , Dongbao Yang , Weiping Wang This is my paper · ORCID

classification cs.CV

keywords curvedtextdetectionweaklysupervisedlearningsemi-supervisedpseudomaskannotationsceneinstancesegmentationrectangleannotationsrecursivetraining

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether accurate curved-text detection, which usually depends on expensive pixel-level polygon annotations, can be trained mostly from cheap rectangle annotations. It proposes a semi- and weakly-supervised framework in which a baseline detector trained on 10% pixel-level annotated images annotates the remaining 90% of weakly annotated or unlabeled images, and a 'local' strategy feeds ground-truth bounding boxes into the mask branch to produce pseudo masks. On CTW1500 and Total-Text, the local strategy reaches 76.0 and 79.4 F-measure, close to the fully supervised upper bound of 78.5 and 81.4 in the same framework and comparable to state-of-the-art methods trained with full annotations. The claim matters because it suggests segmentation-based curved text detectors can be deployed without large pixel-level labeling efforts.

What carries the argument

The load-bearing mechanism is pseudo-mask generation from ground-truth bounding boxes. The fine localization network learns a mapping from region features sampled inside a box to the binary mask of the text instance; when the weakly annotated images supply only horizontal rectangles, those rectangles are fed directly into this network as proposals, and the network's mask output, paired with the ground-truth box, becomes the pseudo annotation used to retrain the detector. In the local strategy, every ground-truth box $g_i$ yields a pseudo mask $m_i$, and $P=\{(g_i,m_i)\mid g_i\in G\}$ is the pseudo annotation set. The box provides localization while the mask branch supplies the curved shape, converting cheap rectangle supervision into dense mask supervision.

What would settle it

Take the weakly annotated images used in the local strategy, compare each generated pseudo mask against withheld polygon ground truth by IoU, and check whether low agreement correlates with the final F-measure gap; if the pseudo masks match the polygons closely, the mechanism is confirmed, and if they are consistently off (missing text tails, including background), the claimed savings would not survive in other settings.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the expensive pixel-level annotation of curved text can be largely replaced by rectangle annotations without losing much accuracy. Using a detector with a coarse localization branch that predicts horizontal boxes and a fine localization branch that predicts text instance masks, the authors train a baseline on 10% pixel-level annotated data, then treat each ground-truth rectangle in the weakly labeled data as a proposal for the fine localization branch and use the predicted mask as a pseudo annotation. Recursive training with these pseudo masks lets the local strategy reach 76.0 F-measure on CTW1500 and 79.4 on Total-Text, versus 78.5 and 81.4 for full supervision, and a 78.1 F-measure in a cross-dataset setting where the model trained on CTW1500 annotates Total-Text. The paper interprets the local strategy's slower saturation across training rounds as evidence that its pseudo masks are cleaner than those produced by naive or filter-based strategies.

Load-bearing premise

The argument depends on the mask branch, trained on just 10% pixel-level data, generalizing to the weakly annotated images well enough that the pseudo masks it produces from ground-truth boxes are accurate rather than systematically wrong.

Editorial extensions

If this is right

With 10% pixel-level annotated data and 90% rectangle-annotated data, the local strategy reaches 76.0 F-measure on CTW1500 and 79.4 on Total-Text, within about 2 points of fully supervised training in the same framework.
The local strategy outperforms the naive and filter strategies by 4.8 and 2.6 F-measure points on CTW1500, respectively, and keeps improving through more training rounds, indicating its pseudo masks carry less noise.
All three strategies beat the baseline trained only on the small pixel-level set, so the framework's recursive self-annotation provides a consistent boost.
In a cross-dataset experiment, the local strategy transfers a detector from CTW1500 to Total-Text using only rectangle annotations on the target set, reaching 78.1 F-measure, close to a leading fully supervised curved-text detector on that benchmark.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the pseudo masks stay accurate when boxes are not tight horizontal rectangles, the same trick could extend to quadrilateral or rotated-box annotations, which would cover more weakly supervised benchmarks; the paper only tests horizontal boxes.
The 10% pixel-level fraction is likely not a magic threshold; a natural extension is to measure how performance degrades as that fraction shrinks toward 1%, which would map the annotation savings more precisely.
The cross-dataset result suggests a practical workflow: keep a pixel-level annotated source set, run the local strategy on any rectangle-annotated target set, and adapt the detector to a new text domain without drawing new polygons.
Directly measuring pseudo-mask quality (for example, IoU against withheld polygon labels) would isolate whether residual errors come from mask generation or from later recursive amplification, a distinction the paper does not test.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The headline claim—comparable with state of the art using only 10% pixel-level annotated data—is plausible but not actually established by the experiments as reported. The local strategy (feed ground-truth boxes into the mask head to synthesize pseudo masks) is a sensible twist on self-training, and the cross-domain CTW1500→Total-Text result is a nice practical bonus. The paper deserves a referee, but the evaluation protocol needs tightening before the central claim can be trusted.

What's genuinely new: the local strategy is not in WeText, Wordsup, or SEE. Instead of thresholding the model's own box/mask outputs, it uses the weak rectangle annotations as proposals and lets the fine localization network produce the mask. That's a reasonable division of labor: the coarse box is trusted, the mask is inferred. The 76.0/79.4 F-measures against fully supervised 78.5/81.4 in the same framework are credible evidence that label efficiency exists. The authors also correctly credit WeText as the closest work and position their contribution as extending it to arbitrary/curved text.

The soft spots are concentrated in the evaluation, and they are load-bearing. The model 'with the top performance across all training rounds is picked for evaluation'—that is test-set selection over the recursive round. Figures 3–4 show local peaks only at round three, so the reported number is the maximum of a small search, not the output of a fixed protocol. Combined with a single random 10%/90% split and no error bars, the gap between Local and Filter/Naive is not clean evidence for the method. The thresholds S, S′, and T are set empirically, with no ablation; if the results are sensitive to them, the robustness story weakens further. The mask-quality concern is real but secondary: if the mask head produces biased masks, recursive training can amplify the bias, and best-round selection would hide exactly that instability. No code is released, which makes re-running the protocol impossible.

The citation pattern is fine; no sign of missing prior work or stacked self-citations. The framework section is clearly written and the figures match the tables.

Who is this for? Anyone working on label-efficient scene text detection, and more broadly on self-training for instance segmentation. As a claim that 10% pixel labels are enough, it needs a multi-split, fixed-round re-run. I would not desk-reject it; I would send it to review and ask for that experiment.

Referee Report

3 major / 6 minor

Summary. The paper proposes a semi- and weakly-supervised framework for curved text detection in natural images. A Mask R-CNN-style detector with a coarse localization (bounding box) branch and a fine localization (mask) branch is first trained on a small amount (10%) of pixel-level annotated data. Three strategies then exploit the remaining 90% of images: a naive semi-supervised strategy that thresholds model outputs, a filter strategy that additionally requires high IoU between detected boxes and ground-truth horizontal rectangles, and a local strategy that feeds ground-truth bounding boxes as proposals to the fine localization network to generate pseudo masks. Recursive training is applied. Experiments on CTW1500 and Total-Text show that the local strategy improves F-measure from 66.1 to 76.0 and from 70.1 to 79.4, respectively, approaching the fully supervised upper bounds of 78.5 and 81.4. A cross-dataset experiment transfers a model from CTW1500 to Total-Text and reports 78.1 F-measure. The central claim is that the method is comparable to state of the art with only 10% pixel-level annotations and 90% rectangle-level weak annotations.

Significance. The paper addresses a practical and timely problem: reducing the annotation cost for curved text detection. The local strategy is simple and the cross-dataset transfer experiment is a valuable addition. If the reported results are robust, the contribution would be a useful step toward low-cost arbitrary-shape text detection. However, the evidence as presented is weakened by evaluation protocol choices (best-round selection on the test set, a single random split with no variance estimate) and by a lack of analysis of the pseudo-mask quality on which the local strategy depends. The paper does not release code or machine-checked proofs, but the described experimental setup is reproducible in principle with additional experiments. Because the core idea is plausible and the reported gains are consistent across two datasets, the contribution merits consideration once the evaluation is strengthened.

major comments (3)

[Section IV.C, Tables I and II] The paper states that 'the model with the top performance across all training rounds is picked for evaluation.' For the three strategies, the reported F-measures are therefore maxima over the three recursive training rounds on the test set, whereas the fully supervised baseline and the SOTA references are single models. This selection procedure inflates the measured improvement (e.g., 76.0 vs 66.1 on CTW1500) and makes the comparison against the fully supervised upper bound (78.5) not a clean estimate of label efficiency. The authors should report the F-measure at every training round, or at least the mean and standard deviation over rounds, and should compare with fully supervised models evaluated under the same best-round selection rule.
[Section IV.A, Tables I and II] For both datasets, a single random 10%/90% split is used (100/900 for CTW1500 and 125/1130 for Total-Text), with no repeated splits or random seeds. Consequently, there is no variance estimate for the headline claim that 10% pixel-level labels plus 90% weak rectangle labels suffice to approach SOTA. The differences between the local strategy (76.0/79.4) and the fully supervised upper bound (78.5/81.4) may be within the noise of the particular split. The authors should repeat the experiments with multiple random splits and report mean and standard deviation.
[Section III-C3, Eq. (4), and Section III-D] The local strategy treats as ground truth the masks produced by the fine localization network from ground-truth rectangles, but this network is trained on only 10% of the images. No analysis is provided of the quality of these pseudo masks (e.g., mask IoU with human annotations on a held-out subset), nor of how masking errors propagate through the recursive training described in Section III-D. Since the reported gains depend directly on the pseudo masks being reliable enough to train from, the authors should include such a quality analysis or an oracle experiment that replaces pseudo masks with ground-truth masks to establish an upper bound.

minor comments (6)

[Section IV.B] The thresholds S (0.5), S' (0.4), and T (0.3) are set empirically without any sensitivity analysis; because they directly control the pseudo-label sets for the naive and filter strategies, a small ablation study would strengthen the paper.
[Section IV.C] The cross-dataset experiment does not state whether the best training round is selected on the Total-Text test set; if so, the same best-round selection issue applies and should be clarified.
[Section III.B] The prediction target for the fine localization network is described only as 'the intersection between an ROI and its associated ground-truth mask'; an explicit equation or figure would improve reproducibility.
[References] Reference [3] contains an author-name typo: 'C. Ya' should be 'C. Yao'.
[Figures 3 and 4] The figures show F-measure versus training round for the three strategies, but the line styles are not defined in a legend, and the y-axis range is not specified; adding a legend and axis range would improve readability.
[Tables I and II] The claim of being 'comparable with state-of-the-art' should be qualified because the cited SOTA methods use different backbones and training data; the fully supervised rows of the same framework provide a fairer reference point.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: pseudo-label self-training is evaluated on held-out benchmarks, so the central claim is not equivalent to its inputs.

full rationale

This paper is an empirical study of semi- and weakly-supervised text detection. The central claim, that the local strategy achieves state-of-the-art-comparable performance with 10% pixel-level annotations and 90% rectangle-level annotations, is supported by F-measure numbers on the held-out CTW1500 and Total-Text test sets. The pseudo annotations generated by the model are not treated as ground truth by definition; whether retraining on them improves held-out performance is an empirically falsifiable question. The local strategy takes ground-truth bounding boxes as proposals and lets the fine localization network produce pseudo masks, but those masks are model outputs whose quality is measured by the final detector on independent test data, not assumed by construction. No load-bearing step reduces to a fitted parameter being renamed as a prediction, and the paper does not rely on self-citations to justify its core premise; its framework is compared against external methods on standard benchmarks. The selection of the best training round for reporting and the use of a single random 10/90 split are methodological robustness concerns, but they are not instances of circular reasoning. Therefore no circularity is identified.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

This is an empirical machine learning paper with no theoretical derivation, so the axiom ledger captures modeling assumptions rather than mathematical axioms. The main free parameters are confidence and IOU thresholds, which are set without sensitivity analysis.

free parameters (4)

score threshold S = 0.5
Used in the naive semi-supervised strategy to accept pseudo annotations; set empirically in Section IV-B with no sensitivity analysis.
score threshold S' = 0.4
Used in the filter strategy; set empirically in Section IV-B without ablation.
IOU threshold T = 0.3
Used in the filter strategy to match detections to ground-truth boxes; set empirically in Section IV-B without ablation.
loss weights lambda1 and lambda2 = 1 and 1
Balancing parameters in Eq. (1); fixed without ablation.

assumptions (3)

domain assumption The fine localization network, trained on 10% pixel-level data, can generate accurate masks from ground-truth boxes on unseen weakly annotated images.
The local strategy in Section III-C3 relies on this assumption to create pseudo masks via Eq. (4).
domain assumption Recursive training with pseudo annotations converges to a better model rather than amplifying label noise.
Section III-D assumes each round's improved model annotates the data more precisely and that the loop improves performance.
domain assumption Ground-truth horizontal rectangles accurately bound the curved text instances.
Used in the filter and local strategies (Sections III-C2 and III-C3) so that IOU filtering and box-based mask generation are meaningful.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Curved Text Detection in Natural Scene Images with Semi- and Weakly-Supervised Learning." pith.science (2026). https://pith.science/paper/RADJRXBH

@misc{pith2026190809990,
  author       = {Pith},
  title        = {Pith review of: Curved Text Detection in Natural Scene Images with Semi- and Weakly-Supervised Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RADJRXBH}},
  note         = {Machine review of arXiv:1908.09990}
}

read the original abstract

Detecting curved text in the wild is very challenging. Recently, most state-of-the-art methods are segmentation based and require pixel-level annotations. We propose a novel scheme to train an accurate text detector using only a small amount of pixel-level annotated data and a large amount of data annotated with rectangles or even unlabeled data. A baseline model is first obtained by training with the pixel-level annotated data and then used to annotate unlabeled or weakly labeled data. A novel strategy which utilizes ground-truth bounding boxes to generate pseudo mask annotations is proposed in weakly-supervised learning. Experimental results on CTW1500 and Total-Text demonstrate that our method can substantially reduce the requirement of pixel-level annotated data. Our method can also generalize well across two datasets. The performance of the proposed method is comparable with the state-of-the-art methods with only 10% pixel-level annotated data and 90% rectangle-level weakly annotated data.

Figures

Figures reproduced from arXiv: 1908.09990 by the authors.

**Figure 2.** Pseudo annotations generation of the local strategy. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** F-measure versus training round for different strategies on [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: F-measure versus training round for different strategies on Total [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

**Figure 5.** Figure 5: Comparison of detection results on CTW1500 in the first row and Total-Text in the last row. Five columns are results of baseline, naive strategy, [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: Annotation results on Total-Text with local strategy using the [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 26 canonical work pages

[1]

Text detection and recognition in imagery: A survey,

Q. Ye and D. Doermann, “Text detection and recognition in imagery: A survey,” TPAMI, vol. 37, no. 7, pp. 1480–1500, 2015

work page 2015
[2]

Scene text detection and recognition: Recent advances and future trends,

Y . Zhu, C. Yao, and X. Bai, “Scene text detection and recognition: Recent advances and future trends,” Frontiers of Computer Science , vol. 10, no. 1, pp. 19–36, 2016

work page 2016
[3]

Scene text detection and recognition: The deep learning era,

S. Long, X. He, and C. Ya, “Scene text detection and recognition: The deep learning era,” arXiv preprint arXiv:1811.04256 , 2018

arXiv 2018
[4]

ICDAR 2013 robust reading competition,

D. Karatzas, F. Shafait, S. Uchida, M. Iwamura, L. G. i Bigorda, S. R. Mestre, J. Mas, D. F. Mota, J. A. Almazan, and L. P. De Las Heras, “ICDAR 2013 robust reading competition,” in ICDAR, 2013, pp. 1484–1493

work page 2013
[5]

ICDAR 2015 competition on robust reading,

D. Karatzas, L. Gomez-Bigorda, A. Nicolaou, S. Ghosh, A. Bagdanov, M. Iwa- mura, J. Matas, L. Neumann, V . R. Chandrasekhar, S. Lu et al. , “ICDAR 2015 competition on robust reading,” in ICDAR, 2015, pp. 1156–1160

work page 2015
[6]

Total-text: A comprehensive dataset for scene text detection and recognition,

C. K. Ch’ng and C. S. Chan, “Total-text: A comprehensive dataset for scene text detection and recognition,” in ICDAR, vol. 1, 2017, pp. 935–942

work page 2017
[7]

Detecting curve text in the wild: New dataset and new solution,

L. Yuliang, J. Lianwen, Z. Shuaitao, and Z. Sheng, “Detecting curve text in the wild: New dataset and new solution,” arXiv preprint arXiv:1712.02170 , 2017

arXiv 2017
[8]

Textsnake: A ﬂexible representation for detecting text of arbitrary shapes,

S. Long, J. Ruan, W. Zhang, X. He, W. Wu, and C. Yao, “Textsnake: A ﬂexible representation for detecting text of arbitrary shapes,” in ECCV, 2018, pp. 19–35

work page 2018

Show all 33 references

[9]

Fused text segmentation networks for multi-oriented scene text detection,

Y . Dai, Z. Huang, Y . Gao, Y . Xu, K. Chen, J. Guo, and W. Qiu, “Fused text segmentation networks for multi-oriented scene text detection,” in ICPR, 2018, pp. 3604–3609

2018
[10]

Mask textspotter: An end-to-end trainable neural network for spotting text with arbitrary shapes,

P. Lyu, M. Liao, C. Yao, W. Wu, and X. Bai, “Mask textspotter: An end-to-end trainable neural network for spotting text with arbitrary shapes,” in ECCV, 2018, pp. 67–83

2018
[11]

Fully convolutional networks for semantic segmentation,

J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in CVPR, 2015, pp. 3431–3440

2015
[12]

Fully convolutional instance-aware semantic segmentation,

Y . Li, H. Qi, J. Dai, X. Ji, and Y . Wei, “Fully convolutional instance-aware semantic segmentation,” in CVPR, 2017, pp. 4438–4446

2017
[13]

Wetext: Scene text detection under weak supervision,

S. Tian, S. Lu, and C. Li, “Wetext: Scene text detection under weak supervision,” in ICCV, 2017, pp. 1492–1500

2017
[14]

Towards end-to-end text spotting with convolu- tional recurrent neural networks,

H. Li, P. Wang, and C. Shen, “Towards end-to-end text spotting with convolu- tional recurrent neural networks,” in ICCV, 2017, pp. 5238–5246

2017
[15]

Learning deconvolution network for semantic segmentation,

H. Noh, S. Hong, and B. Han, “Learning deconvolution network for semantic segmentation,” in ICCV, 2015, pp. 1520–1528

2015
[16]

Mask r-cnn,

K. He, G. Gkioxari, P. Doll ´ar, and R. Girshick, “Mask r-cnn,” in ICCV, 2017, pp. 2980–2988

2017
[17]

Weakly supervised text attention network for generating text proposals in scene images,

R. Li, M. En, J. Li, and H. Zhang, “Weakly supervised text attention network for generating text proposals in scene images,” in ICDAR, vol. 1, 2017, pp. 324–330

2017
[18]

A method for text localization and recognition in real-world images,

L. Neumann and J. Matas, “A method for text localization and recognition in real-world images,” in ACCV, 2010, pp. 770–783

2010
[19]

Text ﬂow: A uniﬁed text detection system in natural scene images,

S. Tian, Y . Pan, C. Huang, S. Lu, K. Yu, and C. Lim Tan, “Text ﬂow: A uniﬁed text detection system in natural scene images,” in ICCV, 2015, pp. 4651–4659

2015
[20]

Wordsup: Exploiting word annotations for character based text detection,

H. Hu, C. Zhang, Y . Luo, Y . Wang, J. Han, and E. Ding, “Wordsup: Exploiting word annotations for character based text detection,” in ICCV, 2017, pp. 4940– 4949

2017
[21]

SEE: towards semi-supervised end-to-end scene text recognition,

C. Bartz, H. Yang, and C. Meinel, “SEE: towards semi-supervised end-to-end scene text recognition,” in AAAI, 2018

2018
[22]

Spatial transformer networks,

M. Jaderberg, K. Simonyan, A. Zisserman et al., “Spatial transformer networks,” in NeurIPS, 2015, pp. 2017–2025

2015
[23]

Deep residual learning for image recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in CVPR, 2016, pp. 770–778

2016
[24]

Feature pyramid networks for object detection,

T.-Y . Lin, P. Doll´ar, R. Girshick, K. He, B. Hariharan, and S. Belongie, “Feature pyramid networks for object detection,” in CVPR, 2017, pp. 2117–2125

2017
[25]

Faster r-cnn: Towards real-time object detection with region proposal networks,

S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,” in NeurIPS, 2015, pp. 91–99

2015
[26]

Fast r-cnn,

R. Girshick, “Fast r-cnn,” in ICCV, 2015, pp. 1440–1448

2015
[27]

mmdetection,

K. Chen, J. Pang, J. Wang, Y . Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Shi, W. Ouyang, C. C. Loy, and D. Lin, “mmdetection,” https://github.com/ open-mmlab/mmdetection, 2018

2018
[28]

Accurate, large minibatch sgd: training imagenet in 1 hour,

P. Goyal, P. Doll ´ar, R. Girshick, P. Noordhuis, L. Wesolowski, A. Kyrola, A. Tulloch, Y . Jia, and K. He, “Accurate, large minibatch sgd: training imagenet in 1 hour,” arXiv preprint arXiv:1706.02677 , 2017

2017 arXiv
[29]

Detecting oriented text in natural images by linking segments,

B. Shi, X. Bai, and S. Belongie, “Detecting oriented text in natural images by linking segments,” arXiv preprint arXiv:1703.06520 , 2017

2017 arXiv
[30]

Detecting text in natural image with connectionist text proposal network,

Z. Tian, W. Huang, T. He, P. He, and Y . Qiao, “Detecting text in natural image with connectionist text proposal network,” in ECCV, 2016, pp. 56–72

2016
[31]

EAST: an efﬁcient and accurate scene text detector,

X. Zhou, C. Yao, H. Wen, Y . Wang, S. Zhou, W. He, and J. Liang, “EAST: an efﬁcient and accurate scene text detector,” in CVPR, 2017, pp. 2642–2651

2017
[32]

Deep matching prior network: Toward tighter multi-oriented text detection,

Y . Liu and L. Jin, “Deep matching prior network: Toward tighter multi-oriented text detection,” in CVPR, 2017, pp. 3454–3461

2017
[33]

Min-entropy latent model for weakly supervised object detection,

F. Wan, P. Wei, Z. Han, J. Jiao, and Q. Ye, “Min-entropy latent model for weakly supervised object detection,” TPAMI, pp. 1–1, 2019

2019

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Text detection and recognition in imagery: A survey,

Q. Ye and D. Doermann, “Text detection and recognition in imagery: A survey,” TPAMI, vol. 37, no. 7, pp. 1480–1500, 2015

work page 2015

[2] [2]

Scene text detection and recognition: Recent advances and future trends,

Y . Zhu, C. Yao, and X. Bai, “Scene text detection and recognition: Recent advances and future trends,” Frontiers of Computer Science , vol. 10, no. 1, pp. 19–36, 2016

work page 2016

[3] [3]

Scene text detection and recognition: The deep learning era,

S. Long, X. He, and C. Ya, “Scene text detection and recognition: The deep learning era,” arXiv preprint arXiv:1811.04256 , 2018

arXiv 2018

[4] [4]

ICDAR 2013 robust reading competition,

D. Karatzas, F. Shafait, S. Uchida, M. Iwamura, L. G. i Bigorda, S. R. Mestre, J. Mas, D. F. Mota, J. A. Almazan, and L. P. De Las Heras, “ICDAR 2013 robust reading competition,” in ICDAR, 2013, pp. 1484–1493

work page 2013

[5] [5]

ICDAR 2015 competition on robust reading,

D. Karatzas, L. Gomez-Bigorda, A. Nicolaou, S. Ghosh, A. Bagdanov, M. Iwa- mura, J. Matas, L. Neumann, V . R. Chandrasekhar, S. Lu et al. , “ICDAR 2015 competition on robust reading,” in ICDAR, 2015, pp. 1156–1160

work page 2015

[6] [6]

Total-text: A comprehensive dataset for scene text detection and recognition,

C. K. Ch’ng and C. S. Chan, “Total-text: A comprehensive dataset for scene text detection and recognition,” in ICDAR, vol. 1, 2017, pp. 935–942

work page 2017

[7] [7]

Detecting curve text in the wild: New dataset and new solution,

L. Yuliang, J. Lianwen, Z. Shuaitao, and Z. Sheng, “Detecting curve text in the wild: New dataset and new solution,” arXiv preprint arXiv:1712.02170 , 2017

arXiv 2017

[8] [8]

Textsnake: A ﬂexible representation for detecting text of arbitrary shapes,

S. Long, J. Ruan, W. Zhang, X. He, W. Wu, and C. Yao, “Textsnake: A ﬂexible representation for detecting text of arbitrary shapes,” in ECCV, 2018, pp. 19–35

work page 2018