A5063778895- Handwritten Text Recognition Techniques
- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
- Image Processing and 3D Reconstruction
- Vehicle License Plate Recognition
- Digital Media Forensic Detection
- Advanced Image Processing Techniques
- Music and Audio Processing
- Hand Gesture Recognition Systems
- Remote-Sensing Image Classification
- Generative Adversarial Networks and Image Synthesis
- Human Pose and Action Recognition
- Diabetic Foot Ulcer Assessment and Management
Chinese Academy of Sciences
2017-2021
University of Chinese Academy of Sciences
2017-2021
Institute of Automation
2018-2021
Beijing Academy of Artificial Intelligence
2021
Shandong Institute of Automation
2019
Reading text in the wild is a challenging task field of computer vision. Existing approaches mainly adopted Connectionist Temporal Classification (CTC) or Attention models based on Recurrent Neural Network (RNN), which computationally expensive and hard to train. In this paper, we present an end-to-end Convolutional for scene recognition. Firstly, instead RNN, adopt stacked convolutional layers effectively capture contextual dependencies input sequence, characterized by lower computational...
Scene text recognition has been widely researched with supervised approaches. Most existing algorithms require a large amount of labeled data and some methods even character-level or pixel-wise supervision information. However, is expensive, unlabeled relatively easy to collect, especially for many languages fewer resources. In this paper, we propose novel semi-supervised method scene recognition. Specifically, design two global metrics, i.e., edit reward embedding reward, evaluate the...
Reading text in the wild is a challenging task computer vision. Scene suffers from various background noise, including shadow, irrelevant symbols and texture. In order to reduce disturbance of we propose dense chained attention network with stacked modules for scene recognition. Each module learns map that adapted corresponding features enhance foreground suppress noise. Besides, branch designed convolution-deconvolution structure which rapidly captures global information guide...
Scene text recognition has received increased attention in the research community. Text wild often possesses irregular arrangements, typically including perspective text, curved oriented text. Most existing methods are hard to work well for especially severely distorted In this paper, we propose a novel Recurrent Calibration Network (RCN) scene recognition. The RCN progressively calibrates boost performance. By decomposing calibration process into multiple steps, can be calibrated normal one...
Images can be considered as the combination of two parts: content and style. The authors’ approach leverage this property by extracting a certain unique style from reference images combining it to generate with new contents. With well‐defined feature extraction module, they propose novel framework various styles same content. To train specific image generation model efficiently, double‐cycle training strategy is proposed: input natural‐content pairs simultaneously, extract their features,...
Scene text recognition has attracted rapidly increasing attention from the research community. Recent dominant approaches typically follow an attention-based encoder-decoder framework that uses a unidirectional decoder to perform decoding in left-to-right manner, but ignoring equally important right-to-left grammar information. In this paper, we propose novel Gate-based Bidirectional Interactive Decoding Network (GBIDN) for scene recognition. Firstly, backward performs right left and...