Arbitrary shape text detection is a challenging task due to the high variety and complexity of scenes texts. In this paper, we propose novel unified relational reasoning graph network for arbitrary detection. our method, an innovative local bridges proposal model via Convolutional Neural Network (CNN) deep Graph (GCN), making end-to-end trainable. To be concrete, every instance will divided into series small rectangular components, geometry attributes (e.g., height, width, orientation)...

Arbitrary shape text detection is a challenging task due to the high complexity and variety of scene texts. In this work, we propose novel adaptive boundary proposal network for arbitrary detection, which can learn directly produce accurate without any post-processing. Our method mainly consists model an innovative deformation model. The constructed by multi-layer dilated convolutions adopted prior information (including classification map, distance field, direction field) coarse proposals....

In arbitrary shape text detection, locating accurate boundaries is challenging and non-trivial. Existing methods often suffer from indirect boundary modeling or complex post-processing. this article, we systematically present a unified coarse-to-fine framework via learning for which can accurately efficiently locate without our method, explicitly model the an innovative iterative transformer in manner. way, method directly gain abandon post-processing to improve efficiency. Specifically,...

Arbitrary shape text detection is a challenging task due to the significantly varied sizes and aspect ratios, arbitrary orientations or shapes, inaccurate annotations, etc. Due scalability of pixel-level prediction, segmentation-based methods can adapt various texts hence attracted considerable attention recently. However, accurate annotations are formidable, existing datasets for scene only provide coarse-grained boundary annotations. Consequently, numerous misclassified pixels background...

Although existing image deep learning super-resolution (SR) methods achieve promising performance on benchmark datasets, they still suffer from severe drops when the degradation of low-resolution (LR) input is not covered in training. To address problem, we propose an innovative unsupervised method Learning Correction Filter via Degradation-Adaptive Regression for Blind Single Image Super-Resolution. Highly inspired by generalized sampling theory, our aims to enhance strength off-the-shelf...

Scene text spotting is a challenging task, especially for inverse-like scene text, which has complex layouts, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">e.g</i> ., mirrored, symmetrical, or retro-flexed. In this paper, we propose unified end-to-end trainable antagonistic framework dubbed IATS, can effectively spot texts without sacrificing general ones. Specifically, an innovative reading-order estimation module (REM) that extracts...

Segmentation-based methods have achieved great success for arbitrary shape text detection. However, separating neighboring instances is still one of the most challenging problems due to complexity texts in scene images. In this article, we propose an innovative kernel proposal network (dubbed KPN) The proposed KPN can separate by classifying different into instance-independent feature maps, meanwhile avoiding complex aggregation process existing segmentation-based detection methods. To be...

Temporal Knowledge Graph (TKG) reasoning often involves completing missing factual elements along the timeline. Although existing methods can learn good embeddings for each element in quadruples by integrating temporal information, they fail to infer evolution of facts. This is mainly because (1) insufficiently exploring internal structure and semantic relationships within individual (2) inadequately learning a unified representation contextual correlations among different quadruples. To...

Arbitrary shape text detection is a challenging task due to the high variety and complexity of scenes texts. In this paper, we propose novel unified relational reasoning graph network for arbitrary detection. our method, an innovative local bridges proposal model via Convolutional Neural Network (CNN) deep Graph (GCN), making end-to-end trainable. To be concrete, every instance will divided into series small rectangular components, geometry attributes (e.g., height, width, orientation)...

Scene text spotting is a challenging task, especially for inverse-like scene text, which has complex layouts, e.g., mirrored, symmetrical, or retro-flexed. In this paper, we propose unified end-to-end trainable antagonistic framework dubbed IATS, can effectively spot texts without sacrificing general ones. Specifically, an innovative reading-order estimation module (REM) that extracts information from the initial boundary generated by (IBM). To optimize and train REM, joint loss consisting...

In arbitrary shape text detection, locating accurate boundaries is challenging and non-trivial. Existing methods often suffer from indirect boundary modeling or complex post-processing. this paper, we systematically present a unified coarse-to-fine framework via learning for which can accurately efficiently locate without our method, explicitly model the an innovative iterative transformer in manner. way, method directly gain abandon post-processing to improve efficiency. Specifically,...

Segmentation-based methods have achieved great success for arbitrary shape text detection. However, separating neighboring instances is still one of the most challenging problems due to complexity texts in scene images. In this paper, we propose an innovative Kernel Proposal Network (dubbed KPN) The proposed KPN can separate by classifying different into instance-independent feature maps, meanwhile avoiding complex aggregation process existing segmentation-based detection methods. To be...

In object detection, non-maximum suppression (NMS) methods are extensively adopted to remove horizontal duplicates of detected dense boxes for generating final instances. However, due the degraded quality detection and not explicit exploration context information, existing NMS via simple intersection-over-union (IoU) metrics tend underperform on multi-oriented long-size objects detection. Distinguishing with general duplicate removal, we propose a novel graph fusion network, named GFNet, Our...

Arbitrary shape text detection is a challenging task due to the significantly varied sizes and aspect ratios, arbitrary orientations or shapes, inaccurate annotations, etc. Due scalability of pixel-level prediction, segmentation-based methods can adapt various texts hence attracted considerable attention recently. However, accurate annotations are formidable, existing datasets for scene only provide coarse-grained boundary annotations. Consequently, numerous misclassified pixels background...

In recent years, attention-based scene text recognition methods have been very popular and attracted the interest of many researchers. Attention-based can adaptively focus attention on a small area or even single point during decoding, in which matrix is nearly one-hot distribution. Furthermore, whole feature maps will be weighted summed by all matrices inference, causing huge redundant computations. this paper, we propose an efficient attention-free Single-Point Decoding Network (dubbed...

Arbitrary shape text detection is a challenging task due to the high complexity and variety of scene texts. In this work, we propose novel adaptive boundary proposal network for arbitrary detection, which can learn directly produce accurate without any post-processing. Our method mainly consists model an innovative deformation model. The constructed by multi-layer dilated convolutions adopted prior information (including classification map, distance field, direction field) coarse proposals....