• Publications
  • Influence
Deep Direct Regression for Multi-oriented Scene Text Detection
TLDR
We analyze the drawbacks of indirect regression, which covers the state-of-the-art detection structures Faster-RCNN and SSD as instances, and point out the potential superiority of direct regression. Expand
  • 209
  • 24
  • PDF
PL-VIO: Tightly-Coupled Monocular Visual–Inertial Odometry Using Point and Line Features
TLDR
This paper proposes point–line visual–inertial odometry (PL-VIO), a tightly-coupled monocular visual– inertial navigation system exploiting both point and line features. Expand
  • 47
  • 8
  • PDF
TextDragon: An End-to-End Framework for Arbitrary Shaped Text Spotting
TLDR
We propose a novel text spotting framework to detect and recognize text of arbitrary shapes in an end-to-end manner, using only word/line-level annotations for training. Expand
  • 29
  • 2
  • PDF
Page Segmentation for Historical Handwritten Documents Using Fully Convolutional Networks
TLDR
We propose a pixel-wise page segmentation method for historical handwritten documents using fully convolutional network (FCN) which yields pixel-level accuracy of above 99%. Expand
  • 21
  • 1
  • PDF
A Unified Video Text Detection Method with Network Flow
TLDR
We propose a novel video text detection method with network flow based tracking. Expand
  • 6
  • 1
Context-aware mathematical expression recognition: An end-to-end framework and a benchmark
TLDR
In this paper we propose a novel end-to-end framework for mathematical expression (ME) recognition, by introducing a CNN based structure with multi-task learning to perform mathematical symbols detection and recognition simultaneously. Expand
  • 10
Realtime multi-scale scene text detection with scale-based region proposal network
TLDR
We propose a two-stage framework for realtime multi-scale scene text detection. Expand
  • 10
Correlational examples for convolutional neural networks to detect small impurities
TLDR
In this paper, a multi-frame detection framework is proposed to directly detect small impurities in opaque glass bottles with liquor using convolutional neural networks. Expand
  • 4
Effects of clay content, cement and mineral composition characteristics on sandstone rock strength and deformability behaviors
Abstract To keep the stability and integrity of wellbore, it is required to figure out the interaction between clays, water and rock strength in the drilling process. As revealed in availableExpand
  • 10
Ensemble Feature Selection for Improving Intrusion Detection Classification Accuracy
TLDR
The idea of ensemble is applied to feature selection to adjust feature subsets to improve classification accuracy of intrusion detection. Expand
  • 1