Pith. sign in

REVIEW 2 major objections 6 minor 59 references

Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network

T0 review · 2 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read A lightweight pipeline detects arbitrary-shaped scene text in real time, matching much heavier detectors.

desk verdict Solid applied text-detection paper whose core claims hold up; d=6 sensitivity and missing contemporaneous comparisons are real but minor conditions, not fatal. read the letter →

arxiv 1908.05900 v2 pith:I2SXTHBN submitted 2019-08-16 cs.CV

classification cs.CV
keywords scenetextdetectionarbitrary-shapedreal-timepixelaggregationfeaturepyramidenhancementsegmentationCTW1500Total-Text
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 argues that arbitrary-shaped text detection does not need a heavy backbone or a slow post-processing stage. It introduces the Pixel Aggregation Network (PAN), built on ResNet-18, a cheap cascaded feature-pyramid enhancement head, and a learnable pixel-aggregation post-process that grows each detected text kernel outward to recover the full text region. On the curved-text benchmarks CTW1500 and Total-Text, PAN-640 reaches 83.7% and 85.0% F-measure respectively at about 40 FPS, while the smaller PAN-320 runs above 80 FPS with only a few points less accuracy. If correct, this makes real-time deployment of curved-text detection feasible on a single GPU.

What carries the argument

The load-bearing mechanism is the learned similarity vector: each text pixel carries a low-dimensional vector (dimension 4), and each text kernel has a centroid vector computed as the mean of its pixels' vectors. Training uses an aggregation loss that pulls pixels toward their own kernel's centroid and a discrimination loss that pushes different kernels' centroids apart. At inference, the detector finds connected components in the predicted kernel map and conditionally absorbs neighboring text pixels whose similarity vectors lie within Euclidean distance $d=6$ of the kernel centroid. The segmentation head supplies the accuracy a lightweight backbone alone lacks: FPEM is a cascadable U-shaped module built from depthwise separable convolutions that fuses low-level and high-level features at about one-fifth the FLOPs of FPN, and FFM combines the enhanced feature pyramids of different depths by element-wise addition followed by upsampling and concatenation.

What would settle it

Build a test grid of tightly packed text lines with known gaps of a few pixels, sweep $d$ from 2 to 10, and count merged or split instances; the paper's calibration claim predicts a wide plateau around $d=6$, while a fragile similarity-vector calibration would show sharp degradation at the default.

Watch

Extended reading notes

Core claim

The central claim is that the usual speed-accuracy trade-off in arbitrary-shaped text detection is avoidable: a ResNet-18 backbone can support competitive detection if the segmentation head is designed to deepen and enlarge receptive fields cheaply (FPEM and FFM), and if the clustering post-process is learned rather than hand-tuned. Pixel Aggregation assigns each text pixel a short learned similarity vector and trains the network so that pixels of one instance sit close to that instance's kernel centroid; at test time, text pixels are merged into a kernel only when the Euclidean distance between their similarity vectors and the kernel's centroid is below a fixed threshold $d=6$. The paper reports that this pipeline reaches 81.0% F-measure on CTW1500 without external data and 83.7% with SynthText pre-training, at 39.8 FPS for 640-pixel input, and 84.2 FPS at 320-pixel input with 79.9% F-measure. The authors state that, to their knowledge, this is the first algorithm that can detect curved text precisely in real time.

Load-bearing premise

The post-processing trusts that a single fixed distance threshold $d=6$ on the learned similarity vectors cleanly separates pixels belonging to a text instance from pixels outside it on every dataset and scale.

Editorial extensions

If this is right

  • Fast curved-text detection becomes practical in a single-GPU, single-thread setting: 84.2 FPS at 320-pixel input and 39.8 FPS at 640-pixel input on CTW1500.
  • A lightweight backbone with the proposed head can match or beat a heavy backbone on segmentation-based text detection: ResNet-18 + 2 FPEMs + FFM reaches nearly the same F-measure as ResNet-50 + PSPNet while running about five times faster.
  • Replacing the naive merge rule with the learned pixel-aggregation step adds more than 1% F-measure on both ICDAR 2015 and CTW1500.
  • The same pipeline handles oriented and long straight text: 82.9% F-measure on ICDAR 2015 and 84.1% on MSRA-TD500 with external pre-training, still above 26 FPS.
  • Cross-dataset tests show the learned representation transfers across datasets: training on CTW1500 and testing on MSRA-TD500 gives 74.8% F-measure, and the reverse direction gives 75.2%.

Reading between the lines

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

  • The learned-clustering recipe could be grafted onto other instance-segmentation tasks where a cheap seed map and a similarity field must separate touching objects, such as cell or panoptic segmentation.
  • The fixed threshold $d=6$ is the least-tested knob; a scale-aware or learned threshold might transfer better across datasets with very different text sizes, so the reported numbers may understate the achievable accuracy on any single dataset.
  • The real-time claim depends on the testing configuration (one 1080Ti GPU, single CPU thread, batch size 1); on weaker hardware the absolute FPS will drop, but the large speed advantage over heavier detectors should persist.
  • Because the cross-dataset results suggest the model learns text shape rather than dataset-specific appearance, a text-spotting system could plausibly use PAN as its front end and inherit the real-time property without changing the recognition stage.
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.

Referee Report

2 major / 6 minor

Summary. This paper proposes Pixel Aggregation Network (PAN), a segmentation-based arbitrary-shaped text detector. The architecture uses a lightweight ResNet18 backbone, a segmentation head composed of cascaded Feature Pyramid Enhancement Modules and a Feature Fusion Module, and a pixel aggregation post-processing step that merges text-region pixels into predicted kernels using learned per-pixel similarity vectors. The method is evaluated on CTW1500, Total-Text, ICDAR 2015, and MSRA-TD500. The main reported results are PAN-640 reaching 83.7% F-measure on CTW1500 (81.0% without external data) at 39.8 FPS, and PAN-320 reaching 79.9% at 84.2 FPS, which the paper claims makes it the first algorithm to detect curved text precisely in real time. Ablations cover the number of FPEMs, the fusion module, the pixel aggregation module, and the backbone choice, and an appendix reports cross-dataset generalization.

Significance. If the reported numbers are correct, the paper makes a valuable practical contribution: it demonstrates that a lightweight segmentation network with a fast aggregation post-processing step can match or exceed the accuracy of heavier arbitrary-shaped text detectors while running an order of magnitude faster. The paper's strengths include systematic ablations (Tables 1-3), fully disclosed hyperparameters, cross-dataset robustness experiments (Table 9), and explicit discussion of failure cases. The pixel aggregation idea, in which the network learns similarity vectors that guide test-time clustering, is simple and potentially useful beyond text detection. The main risks are the lack of sensitivity analysis for the test-time distance threshold and incomplete specification of the aggregation algorithm, both of which affect reproducibility and the generality of the speed-accuracy claim.

major comments (2)
  1. [Sec. 3.4; Sec. 4.2] The post-processing step depends on a fixed Euclidean distance threshold d=6, but the paper reports no sensitivity analysis for this value. The training margins are δ_agg=0.5 and δ_dis=3, so d=6 is not derived from the loss; if the similarity-vector scale varies across datasets or training conditions, the same d could either merge adjacent instances or fragment kernels. Because PA is the paper's core novelty and its test-time behavior is controlled by d, the authors should report F-measure as a function of d (for example d in {2, 4, 6, 8, 10}) on CTW1500 and ICDAR 2015, and ideally on the cross-dataset settings of Appendix 6.1, to demonstrate that the chosen operating point is stable and that the post-processing transfers across datasets.
  2. [Sec. 3.4, step ii] The merging rule is under-specified. It states that a neighbor text pixel is merged when 'the Euclidean distance of their similarity vectors is less than d', but it does not say whether the distance is computed between p and the original kernel center G(K_i), between p and a running mean of the pixels already merged into K_i, or between p and the neighboring pixel through which it is reached, nor whether G(K_i) is updated as pixels are added. These variants can yield different detected instances, especially for large text regions, and prevent exact reproduction. Please specify the exact update rule or release code.
minor comments (6)
  1. [Abstract; Sec. 4.4] The headline number 79.9% at 84.2 FPS on CTW1500 corresponds to PAN-320 with SynthText pre-training; the abstract should state that external data is used to avoid ambiguity.
  2. [Sec. 4.4] The sentence 'PAN-320 ... surpasses most of the counterparts, including the methods with external data pre-training' is misleading because PSENet-1s with external data (82.2) and without external data (78.0) both outperform PAN-320 (77.1) on CTW1500; please rephrase to list which counterparts are surpassed.
  3. [Fig. 1; Sec. 1] The claim that PAN-640 is '10.7% better' than CTD+TLOC is inconsistent with Table 4, where CTD+TLOC has F=73.4 and PAN-640 with external data has F=83.7 (a 10.3-point gap), and without external data has F=81.0 (a 7.6-point gap). Please correct the number or clarify the comparison.
  4. [Sec. 4.4] The ICDAR 2015 paragraph reports a no-external-data F-measure of 80.4%, but Table 6 lists 80.3%; the same value appears as 80.3 in Table 3, so the text should be aligned.
  5. [Sec. 3.4; Sec. 4.2] The term 'learnable post-processing' is slightly overstated because the distance threshold d is manually fixed; consider clarifying that the similarity vectors are learned while d is a hyperparameter.
  6. [Sec. 3.2] The statement that FPEM's FLOPS is 'about 1/5 of FPN' is not supported by any direct comparison; either add the comparison or remove the claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported speed-accuracy results are external benchmark measurements backed by genuine ablations, and the fixed post-processing threshold is a calibration choice rather than a construction.

full rationale

The paper's central claims are empirical comparisons on standard benchmarks (CTW1500, Total-Text, ICDAR 2015, MSRA-TD500), so the F-measure and FPS numbers are not derived from the paper's assumptions by construction. The core novelty, Pixel Aggregation, is tested by a genuine removal experiment: setting beta to 0 in Eq. 5 and merging all neighbor text pixels drops F-measure by over 1%, which is a falsifiable ablation rather than a restatement of the loss. Similarly, FPEM and FFM are ablated by varying the cascade count and comparing fusion methods. The training losses L_agg and L_dis use margins delta_agg=0.5 and delta_dis=3 to shape learned similarity vectors, while the test-time threshold d=6 is a fixed operating point stated in Sec. 4.2; it is not a fitted parameter that defines the evaluation metric, so no prediction reduces to an input fit. The paper contains self-citations (e.g., PSENet for the shrink-ratio ground truth and dice loss, and SPCNet for comparison), but these are standard technique citations and are not load-bearing: the reported results do not reduce to those citations, and the authors' comparison numbers are independently evaluated on public benchmarks. The lack of sensitivity analysis for d is a robustness or calibration concern, not circularity, and the cross-dataset experiments in Appendix 6.1 provide independent generalization evidence. No equation-level reduction of a claimed result to its own inputs was found.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claim is an empirical benchmark result. It depends on standard datasets, pretrained backbones, a set of hand-set hyperparameters (loss weights, margins, kernel shrink ratio, post-processing threshold, cascade depth), and the assumption that pixel similarity vectors generalize across scenes. No new physical or formal entities are introduced.

free parameters (8)
  • alpha (loss weight for kernel loss) = 0.5
    Set experimentally in Eq. 5 and used in all experiments to balance text-region and kernel losses.
  • beta (loss weight for aggregation and discrimination) = 0.25
    Set experimentally in Eq. 5 and used in all experiments to weight the pixel-aggregation losses.
  • delta_agg (margin in aggregation loss) = 0.5
    Set experimentally in Eq. 2 to filter easy samples when pulling text pixels toward their kernel.
  • delta_dis (margin in discrimination loss) = 3
    Set experimentally in Eq. 4 to keep different kernels separated in similarity space.
  • d (post-processing distance threshold) = 6
    Fixed in Sec. 4.2 as the Euclidean distance threshold for merging neighbor text pixels into kernels during testing; no sensitivity analysis is provided.
  • shrink ratio r for kernel ground truth = 0.5 on ICDAR 2015, 0.7 elsewhere
    Chosen per dataset in Sec. 4.2 to generate kernel supervision by shrinking original polygons, following PSENet.
  • number of cascaded FPEMs nc = 2
    Selected by ablation in Sec. 4.3 (Table 1) as the speed-accuracy sweet spot.
  • similarity vector dimension = 4
    Set in Sec. 4.2 as the output dimension of the learned per-pixel similarity vector; no ablation on this dimension is reported.
assumptions (5)
  • domain assumption ImageNet-pretrained backbones transfer to scene text segmentation
    The method relies on ResNet18, ResNet50, and VGG16 pretrained on ImageNet for initialization (Sec. 4.2).
  • domain assumption Shrinking ground-truth polygons by ratio r yields reliable kernel supervision
    Follows PSENet [24]; the kernel representation assumes a single connected shrinking of each text instance is sufficient to separate instances.
  • domain assumption Standard benchmark labels and evaluation protocols are valid measures of text detection quality
    The paper uses CTW1500, Total-Text, ICDAR 2015, and MSRA-TD500 with their standard evaluation procedures (Sec. 4.1).
  • domain assumption Dice loss with OHEM sufficiently handles text/non-text imbalance
    Adopted from prior work [24, 43] and applied in Eq. 6 and Eq. 7 without further justification.
  • domain assumption Fixed Euclidean threshold d=6 in similarity space separates same-instance from different-instance pixel pairs
    Post-processing step ii) in Sec. 3.4 assumes similarity vectors are calibrated so that a single global threshold works across all datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network." pith.science (2026). https://pith.science/paper/I2SXTHBN

@misc{pith2026190805900,
  author       = {Pith},
  title        = {Pith review of: Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/I2SXTHBN}},
  note         = {Machine review of arXiv:1908.05900}
}
read the original abstract

Scene text detection, an important step of scene text reading systems, has witnessed rapid development with convolutional neural networks. Nonetheless, two main challenges still exist and hamper its deployment to real-world applications. The first problem is the trade-off between speed and accuracy. The second one is to model the arbitrary-shaped text instance. Recently, some methods have been proposed to tackle arbitrary-shaped text detection, but they rarely take the speed of the entire pipeline into consideration, which may fall short in practical applications.In this paper, we propose an efficient and accurate arbitrary-shaped text detector, termed Pixel Aggregation Network (PAN), which is equipped with a low computational-cost segmentation head and a learnable post-processing. More specifically, the segmentation head is made up of Feature Pyramid Enhancement Module (FPEM) and Feature Fusion Module (FFM). FPEM is a cascadable U-shaped module, which can introduce multi-level information to guide the better segmentation. FFM can gather the features given by the FPEMs of different depths into a final feature for segmentation. The learnable post-processing is implemented by Pixel Aggregation (PA), which can precisely aggregate text pixels by predicted similarity vectors. Experiments on several standard benchmarks validate the superiority of the proposed PAN. It is worth noting that our method can achieve a competitive F-measure of 79.9% at 84.2 FPS on CTW1500.

Figures

Figures reproduced from arXiv: 1908.05900 by the authors.

Figure 1
Figure 1. The performance and speed on curved text dataset [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall pipeline of PAN. suffer from low inference speed, because of their heavy models or complicated post-processing steps, which lim￾its their deployment in the real-world environment. On the other hand, previous text detectors [58, 32] with high effi￾ciency are mostly designed for quadrangular text instances, which have flaws when detecting curved text. Therefore, “how to design an efficient and accurate arb… view at source ↗
Figure 3
Figure 3. The overall architecture of PAN. The features from lightweight backbone network are enhanced by a low computational-cost [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: The detail of FFM. “+” is element-wise addition. “C” is the operation of upsampling and concatenating. ble convolution, which needs minimal computation. The FLOPS of FPEM is about 1/5 of FPN. 3.3. Feature Fusion Module Feature Fusion Module is applied to fuse the featu…
Figure 6
Figure 6. Figure 6: Qualitative results of PAN. (a) is the final result of PAN. (b) is the predicted text regions. (c) is the predicted kernels. (d) is the [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Failure Samples. of BiSeNet to ResNet18 and use one of default settings of CU-Net, which has the similar speed with our method. As shown in [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Detection results on CTW1500 [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Detection results on Total-Text [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]
Figure 10
Figure 10. Figure 10: Detection results on ICDAR 2015 [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Detection results on MSRA-TD500 [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 48 canonical work pages

  1. [1]

    Rethinking atrous convolution for seman- tic image segmentation

    Liang-Chieh Chen, George Papandreou, Florian Schroff, and Hartwig Adam. Rethinking atrous convolution for seman- tic image segmentation. arXiv preprint arXiv:1706.05587 , 2017

  2. [2]

    Total-text: A com- prehensive dataset for scene text detection and recognition

    Chee Kheng Ch’ng and Chee Seng Chan. Total-text: A com- prehensive dataset for scene text detection and recognition. In Proc. Int. Conf. Document Analysis Recogn., 2017

  3. [3]

    Pix- ellink: Detecting scene text via instance segmentation

    Dan Deng, Haifeng Liu, Xuelong Li, and Deng Cai. Pix- ellink: Detecting scene text via instance segmentation. In Proc. AAAI Conf. Artificial Intell., 2018

  4. [4]

    Imagenet: A large-scale hierarchical im- age database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical im- age database. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2009

  5. [5]

    Salient objects in clutter: Bringing salient object detection to the foreground

    Deng-Ping Fan, Ming-Ming Cheng, Jiang-Jiang Liu, Shang- Hua Gao, Qibin Hou, and Ali Borji. Salient objects in clutter: Bringing salient object detection to the foreground. In Proc. Eur. Conf. Comp. Vis., 2018

  6. [6]

    Structure-measure: A new way to evaluate foreground maps

    Deng-Ping Fan, Ming-Ming Cheng, Yun Liu, Tao Li, and Ali Borji. Structure-measure: A new way to evaluate foreground maps. In Proc. IEEE Int. Conf. Comp. Vis., 2017

  7. [7]

    Enhanced-alignment mea- sure for binary foreground map evaluation

    Deng-Ping Fan, Cheng Gong, Yang Cao, Bo Ren, Ming- Ming Cheng, and Ali Borji. Enhanced-alignment mea- sure for binary foreground map evaluation. arXiv preprint arXiv:1805.10421, 2018

  8. [8]

    Rethinking rgb-d salient object detection: Mod- els, datasets, and large-scale benchmarks

    Deng-Ping Fan, Zheng Lin, Jia-Xing Zhao, Yun Liu, Zhao Zhang, Qibin Hou, Menglong Zhu, and Ming-Ming Cheng. Rethinking rgb-d salient object detection: Mod- els, datasets, and large-scale benchmarks. arXiv preprint arXiv:1907.06781, 2019

Show all 59 references
  1. [9]

    Shifting more attention to video salient ob- ject detection

    Deng-Ping Fan, Wenguan Wang, Ming-Ming Cheng, and Jianbing Shen. Shifting more attention to video salient ob- ject detection. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2019

  2. [10]

    Synthetic data for text localisation in natural images

    Ankush Gupta, Andrea Vedaldi, and Andrew Zisserman. Synthetic data for text localisation in natural images. InProc. IEEE Conf. Comp. Vis. Patt. Recogn., 2016

  3. [11]

    Mask r-cnn

    Kaiming He, Georgia Gkioxari, Piotr Doll ´ar, and Ross Gir- shick. Mask r-cnn. In Proceedings of the IEEE international conference on computer vision, pages 2961–2969, 2017

  4. [12]

    Mask R-CNN

    Kaiming He, Georgia Gkioxari, Piotr Doll ´ar, and Ross Gir- shick. Mask R-CNN. In Proc. IEEE Int. Conf. Comp. Vis., pages 2961–2969, 2017

  5. [13]

    Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification. In Proc. IEEE Int. Conf. Comp. Vis., 2015

  6. [14]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2016

  7. [15]

    Identity mappings in deep residual networks

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Identity mappings in deep residual networks. In Proc. Eur. Conf. Comp. Vis., 2016

  8. [16]

    Single shot text detector with regional attention

    Pan He, Weilin Huang, Tong He, Qile Zhu, Yu Qiao, and Xiaolin Li. Single shot text detector with regional attention. In Proc. IEEE Int. Conf. Comp. Vis., 2017

  9. [17]

    Deep direct regression for multi-oriented scene text detec- tion

    Wenhao He, Xu-Yao Zhang, Fei Yin, and Cheng-Lin Liu. Deep direct regression for multi-oriented scene text detec- tion. In Proc. IEEE Int. Conf. Comp. Vis., 2017

  10. [18]

    Mobilenets: Efficient convolu- tional neural networks for mobile vision applications

    Andrew G Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco An- dreetto, and Hartwig Adam. Mobilenets: Efficient convolu- tional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861, 2017

  11. [19]

    Wordsup: Exploiting word an- notations for character based text detection

    Han Hu, Chengquan Zhang, Yuxuan Luo, Yuzhuo Wang, Junyu Han, and Errui Ding. Wordsup: Exploiting word an- notations for character based text detection. In Proc. IEEE Int. Conf. Comp. Vis., 2017

  12. [20]

    Densely connected convolutional networks

    Gao Huang, Zhuang Liu, Kilian Q Weinberger, and Laurens van der Maaten. Densely connected convolutional networks. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2017

  13. [21]

    Batch normalization: Accelerating deep network training by reducing internal co- variate shift

    Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal co- variate shift. arXiv preprint arXiv:1502.03167, 2015

  14. [22]

    Icdar 2015 competition on ro- bust reading

    Dimosthenis Karatzas, Lluis Gomez-Bigorda, Anguelos Nicolaou, Suman Ghosh, Andrew Bagdanov, Masakazu Iwa- mura, Jiri Matas, Lukas Neumann, Vijay Ramaseshan Chan- drasekhar, Shijian Lu, et al. Icdar 2015 competition on ro- bust reading. In Proc. Int. Conf. Document Analysis Rec...

  15. [23]

    Gradient-based learning applied to document recog- nition

    Yann LeCun, L ´eon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient-based learning applied to document recog- nition. Proceedings of the IEEE, 1998

  16. [24]

    Shape robust text detection with progressive scale expansion network

    Xiang Li, Wenhai Wang, Wenbo Hou, Ruo-Ze Liu, Tong Lu, and Jian Yang. Shape robust text detection with progressive scale expansion network. arXiv preprint arXiv:1806.02559, 2018

  17. [25]

    Selec- tive kernel networks

    Xiang Li, Wenhai Wang, Xiaolin Hu, and Jian Yang. Selec- tive kernel networks. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2019

  18. [26]

    Textboxes++: A single-shot oriented scene text detector

    Minghui Liao, Baoguang Shi, and Xiang Bai. Textboxes++: A single-shot oriented scene text detector. IEEE Trans. Im- age Process., 2018

  19. [27]

    Textboxes: A fast text detector with a single deep neural network

    Minghui Liao, Baoguang Shi, Xiang Bai, Xinggang Wang, and Wenyu Liu. Textboxes: A fast text detector with a single deep neural network. In Proc. AAAI Conf. Artificial Intell. , 2017

  20. [28]

    Rotation-sensitive regression for oriented scene text detection

    Minghui Liao, Zhen Zhu, Baoguang Shi, Gui-song Xia, and Xiang Bai. Rotation-sensitive regression for oriented scene text detection. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2018

  21. [29]

    Feature pyramid networks for object detection

    Tsung-Yi Lin, Piotr Doll ´ar, Ross Girshick, Kaiming He, Bharath Hariharan, and Serge Belongie. Feature pyramid networks for object detection. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2017

  22. [30]

    Ssd: Single shot multibox detector

    Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, and Alexander C Berg. Ssd: Single shot multibox detector. In Proc. Eur. Conf. Comp. Vis., 2016

  23. [31]

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

    Yuliang Liu, Lianwen Jin, Shuaitao Zhang, and Sheng Zhang. Detecting curve text in the wild: New dataset and new solution. 2017

  24. [32]

    Learning markov clustering net- works for scene text detection

    Zichuan Liu, Guosheng Lin, Sheng Yang, Jiashi Feng, Weisi Lin, and Wang Ling Goh. Learning markov clustering net- works for scene text detection. Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2018

  25. [33]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition, pages 3431–3440, 2015

  26. [34]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2015

  27. [35]

    Textsnake: A flexible repre- sentation for detecting text of arbitrary shapes

    Shangbang Long, Jiaqiang Ruan, Wenjie Zhang, Xin He, Wenhao Wu, and Cong Yao. Textsnake: A flexible repre- sentation for detecting text of arbitrary shapes. Proc. Eur. Conf. Comp. Vis., 2018

  28. [36]

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

    Pengyuan Lyu, Minghui Liao, Cong Yao, Wenhao Wu, and Xiang Bai. Mask textspotter: An end-to-end trainable neural network for spotting text with arbitrary shapes. In Proc. Eur. Conf. Comp. Vis., 2018

  29. [37]

    Multi-oriented scene text detection via cor- ner localization and region segmentation

    Pengyuan Lyu, Cong Yao, Wenhao Wu, Shuicheng Yan, and Xiang Bai. Multi-oriented scene text detection via cor- ner localization and region segmentation. arXiv preprint arXiv:1802.08948, 2018

  30. [38]

    Arbitrary-oriented scene text detection via rotation proposals

    Jianqi Ma, Weiyuan Shao, Hao Ye, Li Wang, Hong Wang, Yingbin Zheng, and Xiangyang Xue. Arbitrary-oriented scene text detection via rotation proposals. IEEE Transac- tions on Multimedia, 2018

  31. [39]

    V-net: Fully convolutional neural networks for volumetric medical image segmentation

    Fausto Milletari, Nassir Navab, and Seyed-Ahmad Ahmadi. V-net: Fully convolutional neural networks for volumetric medical image segmentation. In Proc. Int. Conf. 3D Vision, 2016

  32. [40]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. 2017

  33. [41]

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

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In Proc. Advances in Neural Inf. Process. Syst., 2015

  34. [42]

    Detecting oriented text in natural images by linking segments

    Baoguang Shi, Xiang Bai, and Serge Belongie. Detecting oriented text in natural images by linking segments. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2017

  35. [43]

    Training region-based object detectors with online hard ex- ample mining

    Abhinav Shrivastava, Abhinav Gupta, and Ross Girshick. Training region-based object detectors with online hard ex- ample mining. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2016

  36. [44]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. In Proc. Int. Conf. Learn. Representations, 2015

  37. [45]

    On the importance of initialization and momentum in deep learning

    Ilya Sutskever, James Martens, George Dahl, and Geoffrey Hinton. On the importance of initialization and momentum in deep learning. In ICML, 2013

  38. [46]

    Cu-net: Coupled u-nets

    Zhiqiang Tang, Xi Peng, Shijie Geng, Yizhe Zhu, and Dim- itris Metaxas. Cu-net: Coupled u-nets. In BMVC, 2018

  39. [47]

    De- tecting text in natural image with connectionist text proposal network

    Zhi Tian, Weilin Huang, Tong He, Pan He, and Yu Qiao. De- tecting text in natural image with connectionist text proposal network. In Proc. Eur. Conf. Comp. Vis., 2016

  40. [48]

    Mixed link networks

    Wenhai Wang, Xiang Li, Tong Lu, and Jian Yang. Mixed link networks. In Proc. Int. Joint Conf. Artificial Intell., 2018

  41. [49]

    Principal component analysis

    Svante Wold, Kim Esbensen, and Paul Geladi. Principal component analysis. Chemometrics and Intelligent Labora- tory Systems, 2(1-3):37–52, 1987

  42. [50]

    Scene text detection with supervised pyra- mid context network

    Enze Xie, Yuhang Zang, Shuai Shao, Gang Yu, Cong Yao, and Guangyao Li. Scene text detection with supervised pyra- mid context network. In Proc. AAAI Conf. Artificial Intell. , 2019

  43. [51]

    A unified framework for multioriented text detection and recognition.IEEE Trans- actions on Image Processing, 23(11):4737–4749, 2014

    Cong Yao, Xiang Bai, and Wenyu Liu. A unified framework for multioriented text detection and recognition.IEEE Trans- actions on Image Processing, 23(11):4737–4749, 2014

  44. [52]

    Detecting texts of arbitrary orientations in natural images

    Cong Yao, Xiang Bai, Wenyu Liu, Yi Ma, and Zhuowen Tu. Detecting texts of arbitrary orientations in natural images. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2012

  45. [53]

    Scene text detection via holistic, multi-channel prediction

    Cong Yao, Xiang Bai, Nong Sang, Xinyu Zhou, Shuchang Zhou, and Zhimin Cao. Scene text detection via holistic, multi-channel prediction. arXiv preprint arXiv:1606.09002, 2016

  46. [54]

    Bisenet: Bilateral segmentation network for real-time semantic segmentation

    Changqian Yu, Jingbo Wang, Chao Peng, Changxin Gao, Gang Yu, and Nong Sang. Bisenet: Bilateral segmentation network for real-time semantic segmentation. In Proc. Eur. Conf. Comp. Vis., 2018

  47. [55]

    Multi-oriented text detection with fully convolutional networks

    Zheng Zhang, Chengquan Zhang, Wei Shen, Cong Yao, Wenyu Liu, and Xiang Bai. Multi-oriented text detection with fully convolutional networks. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2016

  48. [56]

    Pyramid scene parsing network

    Hengshuang Zhao, Jianping Shi, Xiaojuan Qi, Xiaogang Wang, and Jiaya Jia. Pyramid scene parsing network. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2017

  49. [57]

    Contrast prior and fluid pyramid integration for rgbd salient object detection

    Jia-Xing Zhao, Yang Cao, Deng-Ping Fan, Ming-Ming Cheng, Xuan-Yi Li, and Le Zhang. Contrast prior and fluid pyramid integration for rgbd salient object detection. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2019

  50. [58]

    East: an efficient and ac- curate scene text detector

    Xinyu Zhou, Cong Yao, He Wen, Yuzhi Wang, Shuchang Zhou, Weiran He, and Jiajun Liang. East: an efficient and ac- curate scene text detector. arXiv preprint arXiv:1704.03155, 2017

  51. [59]

    ResNet18 + 2FPEM + FFM

    Appendix 6.1. Robustness Analysis To further demonstrate the robustness of the proposed PAN, we evaluate the model by training on one dataset and testing on other datasets. Based on the annotation level, we divide the datasets into two groups which are word level and text line...

Pith tools

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