A5102008113- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Face and Expression Recognition
- Remote Sensing and Land Use
- Image Enhancement Techniques
- Advanced Image Processing Techniques
- Optical measurement and interference techniques
- Advanced Computational Techniques and Applications
- Evaluation Methods in Various Fields
- Water Quality Monitoring Technologies
- Power Quality and Harmonics
- Robotics and Sensor-Based Localization
- Advanced Algorithms and Applications
- Image and Video Stabilization
- Advanced Image Fusion Techniques
- Video Surveillance and Tracking Methods
- Hydrology and Watershed Management Studies
- Vehicle License Plate Recognition
- Multimodal Machine Learning Applications
- Water Quality and Pollution Assessment
- Impact of Light on Environment and Health
- Automated Road and Building Extraction
- Domain Adaptation and Few-Shot Learning
Henan Polytechnic University
2008-2025
Chongqing University of Posts and Telecommunications
2025
Chongqing Institute of Green and Intelligent Technology
2025
Zhengzhou University of Industrial Technology
2024
Zhengzhou University
2024
Megvii (China)
2021-2022
Vi Technology (United States)
2022
Beijing Institute of Graphic Communication
2019-2021
China Southern Power Grid (China)
2018
Sichuan University
2008-2011
In this paper we present a new data-driven method for pixel-level scene text segmentation from single natural image. Although detection, i.e. producing region mask, has been well studied in the past decade, is still an open problem due to lack of massive labeled data supervised training. To tackle issue, incorporate mask as auxiliary into task, considering acquiring large-scale commonly less expensive and time-consuming. be specific, propose mutually guided network which produces...
We propose a semi-supervised network for wide-angle portraits correction. Wide-angle images often suffer from skew and distortion affected by perspective distortion, especially noticeable at the face regions. Previous deep learning based approaches need ground-truth correction flow maps training guidance. However, such labels are expensive, which can only be obtained manually. In this work, we design scheme build high-quality unlabeled dataset with rich scenarios, allowing us to...
With the growing importance of creative thinking in modern world, there is a need for innovative teaching methods that can effectively cultivate this skill undergraduates. This study introduces an integrated framework combining STEAM (Science, Technology, Engineering, Arts, and Mathematics) education with Problem-Based Learning (PBL) to enhance undergraduates' through linear algebra courses. was implemented during 8-week course at private university Chongqing. incorporated real-world...
Laser space networks are an important development direction for inter-satellite communication. Detecting the angle of arrival (AOA) multiple satellites in a wide field view (FOV) is key to realize laser communication networking. The traditional AOA detection method based on lens system has limited FOV. In this paper, we demonstrate that uses spatially distributed sensor array detect basic concept using signal strength each at different spatial angles. An model was developed, and relationship...
Wide-angle portraits often enjoy expanded views. However, they contain perspective distortions, especially noticeable when capturing group portrait photos, where the background is skewed and faces are stretched. This paper introduces first deep learning based approach to remove such artifacts from freely-shot photos. Specifically, given a wide-angle as input, we build cascaded network consisting of LineNet, ShapeNet, transition module (TM), which corrects distortions on background, adapts...
The development and implementation of an innovative system designed to automatically score English writing ability using advanced machine learning algorithms is challenging task. core objective the study establish a reliable efficient method for assessing written English, which crucial in educational professional settings. paper begins with overview existing methods assessment, highlighting their limitations, such as time consumption potential biases human evaluation. main focus design...
Research on soil organic carbon (SOC) is crucial for improving sinks and achieving the "double-carbon" goal. This study introduces ten auxiliary variables based data from a 2021 land quality survey in Zhengzhou multi-objective regional geochemical survey. It uses geostatistical ordinary kriging (OK) interpolation, as well classical machine learning (ML) models, including random forest (RF) support vector (SVM), to map density (SOCD) topsoil layer (0 − 20 cm) of cultivated land. partitions...
A coding metasurface is designed based on hybrid Array pattern synthesis (APS) and Particle swarm optimization (PSO) method for ultra-wideband low-detectable application. The composed of Electromagnetic band-gap (EBG) structures two Minkowski fractal elements with reflection phase difference 180° (±37°) over a wide frequency range. Two different types EBG unit cell printed thin grounded dielectric substrate produce about 180 degrees Ultra-wideband Radar cross section (RCS) reduction results...
With the rapid development of Internet and update iteration electronic products, augmented reality technology is widely used in military, medical, education, entertainment other fields. Compared with traditional AR technology, WebAR can efficiently run on web browsers various systems be without additional APP. achieves advantages a light-weight, easy spread cross-platform. these digital means, way human-computer interaction presentation also changing each passing day. It gradually being...
Abstract Recently, owing to the requirements of inference speed, most real-time semantic segmentation networks often have shallow network depth, which limits receptive field size model, leading limited acquisition information and resulting in intraclass inconsistency ultimately a decrease accuracy. Additionally, depth also restricts feature extraction capability network, reducing its robustness ability adapt complex scenes. To address these issues, bilateral with rich extractor (RSE) for...
With the continuous development of e-commerce, consumers show increasing interest in posting comments on consumption experience and quality commodities. Meanwhile, people make purchasing decisions relying other much more than ever before. So reliability commodity has a significant impact ensuring consumers’ equity building fair internet-trade-environment. However, some unscrupulous online-sellers write fake praiseful reviews for themselves malicious their business counterparts to maximize...
By efficiently and accurately predicting the adoptability of pets, shelters rescuers can be positively guided on improving attraction pet profiles, reducing animal suffering euthanization. Previous prediction m... | Find, read cite all research you need Tech Science Press
This paper proposes a high-performance resonant current controller for shunt active power filter (APF). The method employs an array of controllers implemented in fundamental reference frame to selective harmonic compensation (SHC). based on the pole-zero cancellation technique, taking into account control object transfer function at each frequency. To cope with grid frequency variations, utilizes information provided by Phase-Locked Loop (PLL). Furthermore, relationship between parameters...
Semantic segmentation based on deep learning has undergone remarkable advancements in recent years. However, due to the neglect of shallow features, problems inaccurate have persisted. To address this issue, a semantic network-attention-based auxiliary extraction and hybrid subsampled network (AAEHS-Net) is suggested study. extract more information complementary enhanced module (CEEM) utilized by network. As result, edge model improved. Moreover, reduce loss (HSM) introduced. Meanwhile,...
Multimodal entity linking (MEL) aims to utilize multimodal information (usually textual and visual information) link ambiguous mentions unambiguous entities in knowledge base. Current methods facing main issues: (1)treating the entire image as input may contain redundant information. (2)the insufficient utilization of entity-related information, such attributes images. (3)semantic inconsistency between base its representation. To this end, we propose DWE+ for linking. could capture finer...
Based on the instantaneous reactive power theory, an improving detection method of harmonic current is put forward. In this method, detected by average. The has characteristics simple structure, strong real-time and good dynamic response. paper, proposed used to carry out simulation MATLAB results show effectiveness method.
Salient point is an important feature in the content-based image retrieval. A novel retrieval based on salient proposed this paper. Firstly, improved algorithm presented SPARSE(Salient Points Auto-Reduction using Segmentation), which dynamic threshold segmentation between-class variance and within-class of segmentation. And then three color features texture are adopted by Experimental results show that method has better performance.
An image retrieval based on texture primitive is proposed in this paper. The firstly constructed according to the principle of block truncation code (BTC). Meantime, analysis statistical distribution primitive, spatial map each feature presented describe information. Experimental results show that method has sound and robust performance by integrating information into descriptors.