A5102810655- Handwritten Text Recognition Techniques
- Vehicle License Plate Recognition
- Image Processing and 3D Reconstruction
- Advanced Image and Video Retrieval Techniques
- Digital Media Forensic Detection
- Generative Adversarial Networks and Image Synthesis
- Text and Document Classification Technologies
- Natural Language Processing Techniques
- Web Data Mining and Analysis
- Topic Modeling
- Plant Stress Responses and Tolerance
- Multimodal Machine Learning Applications
- Software Engineering Research
- Photoacoustic and Ultrasonic Imaging
- Plant Molecular Biology Research
- Image and Object Detection Techniques
- Plant Genetic and Mutation Studies
- Image Enhancement Techniques
- Seed Germination and Physiology
- Advanced Vision and Imaging
- Advanced Image Processing Techniques
- Plant responses to elevated CO2
Henan Agricultural University
2023-2025
Harbin University of Science and Technology
2023
South China University of Technology
2018-2021
Curved text detection is a difficult problem that has not been addressed sufficiently. To highlight the difficulties in reading curved real environment, we constructed dataset called CTW1500, which includes over 10,000 annotations 1500 images, and used it to formulate polygon-based detector can detect without using an empirical combination. With seamless integration of recurrent transverse longitudinal offset connection, our method explores context information instead predicting points...
Scene text detection has been made great progress in recent years. The manners are evolving from axis-aligned rectangle to rotated and further quadrangle. However, current datasets contain very little curve text, which can be widely observed scene images such as signboard, product name so on. To raise the concerns of reading wild, this paper, we construct a dataset named CTW1500, includes over 10k annotations 1,500 (1000 for training 500 testing). Based on dataset, pioneering propose polygon...
This paper reports the ICDAR2019 Robust Reading Challenge on Arbitrary-Shaped Text - RRC-ArT that consists of three major challenges: i) scene text detection, ii) recognition, and iii) spotting. A total 78 submissions from 46 unique teams/individuals were received for this competition. The top performing score each challenge is as follows: T1 82.65%, T2.1 74.3%, T2.2 85.32%, iv) T3.1 53.86%, v) T3.2 54.91%. Apart results, also details ArT dataset, tasks description, evaluation metrics...
A new method is proposed for removing text from natural images. The challenge to first accurately localize on the stroke-level and then replace it with a visually plausible background. Unlike previous methods that require image patches erase scene text, our method, namely ensconce network (EnsNet), can operate end-to-end single without any prior knowledge. overall structure an trainable FCN-ResNet-18 conditional generative adversarial (cGAN). feature of former enhanced by novel lateral...
Scene text removal has attracted increasing research interests owing to its valuable applications in privacy protection, camera-based virtual reality translation, and image editing. However, existing approaches, which fall short on real applications, are mainly because they were evaluated synthetic or unrepresentative datasets. To fill this gap facilitate direction, article proposes a real-world dataset called SCUT-EnsText that consists of 3,562 diverse images selected from public scene...
Visual Information Extraction (VIE) has attracted considerable attention recently owing to its various advanced applications such as document understanding, automatic marking and intelligent education. Most existing works decoupled this problem into several independent sub-tasks of text spotting (text detection recognition) information extraction, which completely ignored the high correlation among them during optimization. In paper, we propose a robust System (VIES) towards real-world...
Low temperatures disrupt nitrogen metabolism in tobacco, resulting lower nicotine content the leaves. 24-epibrassinolide (EBR) is a widely used plant growth regulator known for its roles enhancing cold tolerance and metabolism. Nevertheless, it remains unclear whether EBR enhances leaf under low temperature conditions during mature stage of flue-cured tobacco. To investigate effects on 'Yunyan 87' four treatments (foliar spraying 0, 0.1, 0.2 0.4 mg·L− 1 solutions) were performed by using...
Evaluation protocols play key role in the developmental progress of text detection methods. There are strict requirements to ensure that evaluation methods fair, objective and reasonable. However, existing metrics exhibit some obvious drawbacks: 1) They not goal-oriented; 2) they cannot recognize tightness methods; 3) one-to-many many-to-one solutions involve inherent loopholes deficiencies. Therefore, this paper proposes a novel protocol called Tightness-aware Intersect-over-Union (TIoU)...
This paper reports the ICDAR2019 Robust Reading Challenge on Arbitrary-Shaped Text (RRC-ArT) that consists of three major challenges: i) scene text detection, ii) recognition, and iii) spotting. A total 78 submissions from 46 unique teams/individuals were received for this competition. The top performing score each challenge is as follows: T1 - 82.65%, T2.1 74.3%, T2.2 85.32%, iv) T3.1 53.86%, v) T3.2 54.91%. Apart results, also details ArT dataset, tasks description, evaluation metrics...
To date, offline handwriting paragraph recognition systems either separately crop text line images and recognize them or perform implicit segmentation by integrating complicated multi-dimensional long short-term memory (MDLSTM) with an attention mechanism. The former abovementioned approachs could lead to sub-optimal performances while the latter is very time-consuming. In this paper, a fast end-to-end system, called adversarial feature enhancing network (AFEN), proposed for handwritten...
Multi-orientation scene text detection has recently gained significant research attention. Previous methods directly predict words or lines, typically by using quadrilateral shapes. However, many of these neglect the significance consistent labeling, which is important for maintaining a stable training process, especially when it comprises large amount data. Here we solve this problem proposing new method, Orderless Box Discretization (OBD), first discretizes box into several key edges...
A new method is proposed for removing text from natural images. The challenge to first accurately localize on the stroke-level and then replace it with a visually plausible background. Unlike previous methods that require image patches erase scene text, our method, namely ensconce network (EnsNet), can operate end-to-end single without any prior knowledge. overall structure an trainable FCN-ResNet-18 conditional generative adversarial (cGAN). feature of former enhanced by novel lateral...
Evaluation protocols play key role in the developmental progress of text detection methods. There are strict requirements to ensure that evaluation methods fair, objective and reasonable. However, existing metrics exhibit some obvious drawbacks: 1) They not goal-oriented; 2) they cannot recognize tightness methods; 3) one-to-many many-to-one solutions involve inherent loopholes deficiencies. Therefore, this paper proposes a novel protocol called Tightness-aware Intersect-over-Union (TIoU)...
In array ultrasound imaging, because the software needs to process a large amount of raw data, it is not easy improve imaging rate system. This paper uses FPGA realize high-speed processing images based on transducers with 64 signals intercepted in parallel. After that, performs histogram equalization image FPGA, and finally sends processed result computer for display. study can significantly signal rate, thereby increasing continuous frame addition, when designing pixel value accumulation...