Text-Attentional Convolutional Neural Network for Scene Text Detection

By Tong He, Weilin Huang, Yu Qiao, Jian Yao. IEEE Transactions on Image Processing, 2015

321 citations recorded by the source corpus.

Recent deep learning models have demonstrated strong capabilities for classifying text and non-text components in natural images. They extract a high-level feature globally computed from a whole image component (patch), where the cluttered background information may dominate true text features in the deep representation. This leads to less discriminative power and poorer robustness. In this paper, we present a new system for scene text detection by proposing a novel text-attentional convolutional neural network (Text-CNN) that particularly focuses on extracting text-related regions and features from the image components. We develop a new learning mechanism to train the Text-CNN with multi-level and rich supervised information, including text region mask, character label, and binary text/non-text information. The rich supervision information enables the Text-CNN with a strong capability for discriminating ambiguous texts, and also increases its robustness against complicated background components. The training process is formulated as a multi-task learning problem, where low-level supervised information greatly facilitates the main task of text/non-text classification. In addition, a powerful low-level detector called contrast-enhancement maximally stable extremal regions (MSERs) is developed, which extends the widely used MSERs by enhancing intensity contrast between text patterns and background. This allows it to detect highly challenging text patterns, resulting in a higher recall. Our approach achieved promising results on the ICDAR 2013 data set, with an F-measure of 0.82, substantially improving the state-of-the-art results.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/d0686e9e1d3096b80e9ce192c497a71d672fbed5

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. Yu Qiao
5. Dacheng Tao
6. Luc Van Gool
7. Minghui Liao
8. Kai Chen

In Media Technology

1. Cong Yao
2. Xiang Bai
3. Minghui Liao
4. Lianwen Jin
5. Zhaoyi Wan
6. Yuliang Liu
7. Tong He
8. Kai Chen

Citations in the corpus

• CNN-RNN Based Intelligent Recommendation for Online Medical Pre-Diagnosis Support (2020, methodology): “He et al. [19] constructed a text-attentional convolutional neural network model, which aimed to extract the text-related region and specific text characteristics from a given image component, using multi-level supervised information.”
• Arbitrary-Oriented Scene Text Detection via Rotation Proposals (2017, methodology): “Recently, convolution-network-based approaches were proposed to perform text detection, e.g., Text-CNN [38], by first using an optimized MSER detector to find the approximate region of the text and then sending region features into a characterbased horizontal text CNN classifier to further recognize”
• Object Detection in 20 Years: A Survey (2019, methodology): “The first group of methods frame the text detection as a special case of general object detection [251, 348–357].”
• Deep Direct Regression for Multi-oriented Scene Text Detection (2017, unknown): “[6] 0.”
• ABCNet: Real-Time Scene Text Spotting With Adaptive Bezier-Curve Network (2020, background): “Despite tremendous progress has been made recently [10, 41, 27, 35, 26, 42], detecting and recognizing text in the wild remains largely unsolved due to its diversity patterns in sizes, aspect ratios, font styles, perspective distortion, and shapes.”
• An End-to-End TextSpotter with Explicit Alignment and Attention (2018, unknown): “Recent approaches cast previous character-based text detection [14, 15, 11, 37] into direct text region estimation [24, 43, 42, 10, 8, 41, 38], which avoids multiple bottom-up post-processing steps by taking word or text-line as a whole detection unite.”
• TextBoxes++: A Single-Shot Oriented Scene Text Detector (2018, unknown): “tEval Methods R P F fasttext [54] 0.69 0.84 0.77 – – – MMser [55] 0.70 0.86 0.77 – – – Lu et al. [56] 0.70 0.89 0.78 – – – TextFlow [57] 0.76 0.85 0.80 – He et al. [58] 0.76 0.85 0.80 – – – He et al. [59] 0.73 0.93 0.82 – – – FCRNall+filts [27] – – – 0.76 0.92 0.83 FCN [30] 0.78 0.88 0.83 – Tian et a”
• Detecting Text in Natural Image with Connectionist Text Proposal Network (2016, background): “ing the F-measure from 0.80 to 0.88. The gains are considerable in both precision and recall, with more than +5% and +7% improvements, respectively. In addition, we further compare our method against [8,11,35], which were published after our initial submission. It consistently obtains substantial im”
• Single Shot Text Detector with Regional Attention (2017, methodology): “.69 0.84 0.77 - - - 0.15 CASIA USTB 0.40 0.62 0.48 Zhang [35] 0.74 0.88 0.80 0.76 0.88 0.82 60 StradVision1 0.46 0.53 0.50 TextFlow [27] 0.76 0.85 0.80 - - - 0.94 StradVision2 0.37 0.77 0.50 Text-CNN [8] 0.73 0.93 0.82 0.76 0.93 0.84 4.6 MCLAB FCN [36] 0.43 0.71 0.54 FCRN [4] 0.76 0.94 0.84 0.76 0.9”
• Real-time Scene Text Detection with Differentiable Binarization (2019, methodology): “(He et al. 2016b) 71 61 69 DeepReg (He et al.”
• A Fast Uyghur Text Detector for Complex Background Images (2018, methodology): “At the same time, whether the window is a text region or not, it can be distinguished with a classifier, such as AdaBoost [13], [14], random forest [15], support vector machine (SVM) [16] or convolutional neural network (CNN) [17]–[19].”