Multi-oriented Text Detection with Fully Convolutional Networks

By Zheng Zhang, Chengquan Zhang, Wei Shen, Cong Yao, Xiang Bai. Computer Vision and Pattern Recognition, 2016

534 citations recorded by the source corpus.

In this paper, we propose a novel approach for text detection in natural images. Both local and global cues are taken into account for localizing text lines in a coarse-to-fine procedure. First, a Fully Convolutional Network (FCN) model is trained to predict the salient map of text regions in a holistic manner. Then, text line hypotheses are estimated by combining the salient map and character components. Finally, another FCN classifier is used to predict the centroid of each character, in order to remove the false hypotheses. The framework is general for handling text in multiple orientations, languages and fonts. The proposed method consistently achieves the state-of-the-art performance on three text detection benchmarks: MSRA-TD500, ICDAR2015 and ICDAR2013.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/f740d68440e1a2698d89ee36b21358f4d2c8c1b7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Xiang Bai
2. Chunhua Shen
3. Cong Yao
4. Luc Van Gool
5. Dacheng Tao
6. Yu Qiao
7. Xiaogang Wang
8. Alan L. Yuille

In Media Technology

1. Cong Yao
2. Xiang Bai
3. Minghui Liao
4. Xin He
5. Shangbang Long
6. Chunhua Shen
7. Pengyuan Lyu
8. Lianwen Jin

Citations in the corpus

• Deep Direct Regression for Multi-oriented Scene Text Detection (2017, methodology): “FCNText [26] adopts the FCN [16] for object segmentation to segment the text region by a coarse-to-fine process.”
• FoveaBox: Beyond Anchor-based Object Detector (2019, methodology): “The score mask technique has been widely used in the area of text detection [48][18][50].”
• TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes (2018, methodology): “[46] first use FCN to extract text blocks and then hunt character candidates from these blocks with MSER [24].”
• Look More Than Once: An Accurate Detector for Text of Arbitrary Shapes (2019, methodology): “[40] first used FCN [20] to extract text blocks and then hunted text lines with the statistical information of MSERs [27].”
• Robust Scene Text Recognition with Automatic Rectification (2016, methodology): “In the future, we plan to address the end-to-end scene text reading problem through the combination of RARE with a scene text detection method, e.g. [43].”
• Recognizing Text-Based Traffic Guide Panels with Cascaded Localization Network (2016, unknown): “Most state-of-the-art methods of scene text extraction [1,2,4,5,6,7,8,9,10] comprise two stages, detection to obtain image regions containing text information, and recognition to transform image-based text information into text codes.”
• Rotation-Sensitive Regression for Oriented Scene Text Detection (2018, background): “The works in [11, 26] are such examples; 2) Multi-oriented text detectors [48, 50, 51, 38, 29, 13, 12, 14, 46] which focus on detecting text of arbitrary orientations.”
• Total-Text: A Comprehensive Dataset for Scene Text Detection and Recognition (2017, methodology): “Similar to [9], [10], we first generated the binary mask with 1 indicating text region and 0 for background.”
• Sliding Line Point Regression for Shape Robust Scene Text Detection (2018, unknown): “[18] trained a fully convolutional network (FCN) [19], [20] to predict the salient map of text regions, then traced the text line by combining the salient map and character components.”
• Character Region Awareness for Text Detection (2019, methodology): “Character-level text detectors Zhang et al. [39] proposed a character level detector using text block candidates distilled by MSER [27].”
• FoveaBox: Beyound Anchor-Based Object Detection (2019, methodology): “Besides, more significantly, these methods are mainly used in special domain objection detection such as scene text detection [17], [18] and face detection [19].”
• Efficient and Accurate Arbitrary-Shaped Text Detection With Pixel Aggregation Network (2019, methodology): “[54] first estimatetext blocks with FCNs and detectcharacter candidates from those text blocks with MSER.”
• Gliding Vertex on the Horizontal Bounding Box for Multi-Oriented Object Detection (2019, methodology): “The mainstream CNN-based detectors can be roughly divided into regression-based and segmentation-based [12], [29] methods.”
• Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection (2017, methodology): “…in [10], Huang et al . integrated MSER and CNN to significantly enhance performance over conventional methods; Zhang et al . utilized Fully Convolutional Network [39] to efficiently generate a pixel-wise text/non-text salient map, which achieve state-of-the-art performance on public datasets.”
• OCR-Free Document Understanding Transformer (2021, methodology): “Early detection methods used CNNs to predict local segments and apply heuristics to merge them [19,69].”
• Single Shot Text Detector with Regional Attention (2017, methodology): “rt results in [18,28]. Demo is available at: http://sstd.whuang.org/. 1. Introduction Reading text in the wild has attracted increasing attention in computer vision community, as shown in recent work [28,4,36,38,2,6]. It has numerous potential applications in image retrieval, industrial automation, ”
• Shape Robust Text Detection With Progressive Scale Expansion Network (2018, methodology): “[40] first adopted FCN to extract text blocks and detect character candidates from those text blocks via MSER.”