TextField: Learning A Deep Direction Field for Irregular Scene Text Detection
arXiv:1812.01393 · doi:10.1109/TIP.2019.2900589
Abstract
Scene text detection is an important step of scene text reading system. The main challenges lie on significantly varied sizes and aspect ratios, arbitrary orientations and shapes. Driven by recent progress in deep learning, impressive performances have been achieved for multi-oriented text detection. Yet, the performance drops dramatically in detecting curved texts due to the limited text representation (e.g., horizontal bounding boxes, rotated rectangles, or quadrilaterals). It is of great interest to detect curved texts, which are actually very common in natural scenes. In this paper, we present a novel text detector named TextField for detecting irregular scene texts. Specifically, we learn a direction field pointing away from the nearest text boundary to each text point. This direction field is represented by an image of two-dimensional vectors and learned via a fully convolutional neural network. It encodes both binary text mask and direction information used to separate adjacent text instances, which is challenging for classical segmentation-based approaches. Based on the learned direction field, we apply a simple yet effective morphological-based post-processing to achieve the final detection. Experimental results show that the proposed TextField outperforms the state-of-the-art methods by a large margin (28% and 8%) on two curved text datasets: Total-Text and CTW1500, respectively, and also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500. Furthermore, TextField is robust in generalizing to unseen datasets. The code is available at https://github.com/YukangWang/TextField.
To appear in IEEE TIP
References in corpus (11)
- Adam: A Method for Stochastic Optimization
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Caffe: Convolutional Architecture for Fast Feature Embedding
- Fully Convolutional Networks for Semantic Segmentation
- Arbitrary-Oriented Scene Text Detection via Rotation Proposals
- TextBoxes++: A Single-Shot Oriented Scene Text Detector
- TextBoxes: A Fast Text Detector with a Single Deep Neural Network
- Text-Attentional Convolutional Neural Networks for Scene Text Detection
- Detecting Curve Text in the Wild: New Dataset and New Solution
- Scene Text Detection via Holistic, Multi-Channel Prediction
- DeepFlux for Skeletons in the Wild
Cited by in corpus (47)
- Gliding vertex on the horizontal bounding box for multi-oriented object detection
- Adaptive Period Embedding for Representing Oriented Objects in Aerial Images
- A Single-Shot Arbitrarily-Shaped Text Detector based on Context Attended Multi-Task Learning
- Real-time Scene Text Detection with Differentiable Binarization
- Text Recognition in the Wild: A Survey
- ABCNet: Real-time Scene Text Spotting with Adaptive Bezier-Curve Network
- CT-Net: Arbitrary-Shaped Text Detection via Contour Transformer
- TextDCT: Arbitrary-Shaped Text Detection via Discrete Cosine Transform Mask
- Text Detection and Recognition in the Wild: A Review
- Fourier Contour Embedding for Arbitrary-Shaped Text Detection
- PAN++: Towards Efficient and Accurate End-to-End Spotting of Arbitrarily-Shaped Text
- Towards Unconstrained End-to-End Text Spotting
- ContourNet: Taking a Further Step toward Accurate Arbitrary-shaped Scene Text Detection
- Deep Relational Reasoning Graph Network for Arbitrary Shape Text Detection
- ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
- CBNet: A Plug-and-Play Network for Segmentation-Based Scene Text Detection
- CentripetalText: An Efficient Text Instance Representation for Scene Text Detection
- Arbitrary-Shaped Text Detection withAdaptive Text Region Representation
- Learning Directional Feature Maps for Cardiac MRI Segmentation
- PolarMask++: Enhanced Polar Representation for Single-Shot Instance Segmentation and Beyond
- Convolutional Character Networks
- FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
- FC2RN: A Fully Convolutional Corner Refinement Network for Accurate Multi-Oriented Scene Text Detection
- TextCohesion: Detecting Text for Arbitrary Shapes
- Image Segmentation For Improved Lossless Screen Content Compression
- Mask is All You Need: Rethinking Mask R-CNN for Dense and Arbitrary-Shaped Scene Text Detection
- Weakly-Supervised Arbitrary-Shaped Text Detection with Expectation-Maximization Algorithm
- MOST: A Multi-Oriented Scene Text Detector with Localization Refinement
- ReLaText: Exploiting Visual Relationships for Arbitrary-Shaped Scene Text Detection with Graph Convolutional Networks
- Self-Training for Domain Adaptive Scene Text Detection
- Text Perceptron: Towards End-to-End Arbitrary-Shaped Text Spotting
- Synthetic-to-Real Unsupervised Domain Adaptation for Scene Text Detection in the Wild
- Exploring the Capacity of an Orderless Box Discretization Network for Multi-orientation Scene Text Detection
- Character Region Attention For Text Spotting
- EATEN: Entity-aware Attention for Single Shot Visual Text Extraction
- TextRay: Contour-based Geometric Modeling for Arbitrary-shaped Scene Text Detection
- Arbitrarily Shaped Scene Text Detection: A Decade of Advances and Systematic Analysis
- Scene Text Detection with Scribble Lines
- Box2Poly: Memory-Efficient Polygon Prediction of Arbitrarily Shaped and Rotated Text
- MT: Multi-Perspective Feature Learning Network for Scene Text Detection
- DGST : Discriminator Guided Scene Text detector
- All You Need is a Second Look: Towards Arbitrary-Shaped Text Detection
- Attention-based Feature Decomposition-Reconstruction Network for Scene Text Detection
- Adaptive Shrink-Mask for Text Detection
- Bidirectional Regression for Arbitrary-Shaped Text Detection
- All you need is a second look: Towards Tighter Arbitrary shape text detection
- A method for detecting text of arbitrary shapes in natural scenes that improves text spotting