A5014484224- Remote Sensing and LiDAR Applications
- 3D Surveying and Cultural Heritage
- Advanced Image and Video Retrieval Techniques
- Image Processing and 3D Reconstruction
- Advanced Neural Network Applications
- Robotics and Sensor-Based Localization
- Network Security and Intrusion Detection
- Remote-Sensing Image Classification
- Advanced Vision and Imaging
- Generative Adversarial Networks and Image Synthesis
- Remote Sensing and Land Use
- Automated Road and Building Extraction
- Big Data Technologies and Applications
- 3D Shape Modeling and Analysis
- Diamond and Carbon-based Materials Research
- Augmented Reality Applications
- Advanced Graph Neural Networks
- Vehicle License Plate Recognition
- Image Enhancement Techniques
- Lubricants and Their Additives
- Medical Image Segmentation Techniques
- Ferroelectric and Negative Capacitance Devices
- Phase-change materials and chalcogenides
- AI and Big Data Applications
- Model Reduction and Neural Networks
Sanya University
2020-2025
Beijing Normal University
2015
Recovering a complete point cloud from partial is critical and challenging task for many 3D applications. In this paper, completion network proposed that focuses on improving the feature extraction initial generated in encoding phase. The local details of original are highlighted by introducing trigonometric positional embedding encoding. Moreover, self-attention mechanism fusion to facilitate generation cloud. experiments multiple public datasets demonstrate our effectively achieves with...
Semantic scene completion through AI-driven content generation is a rapidly evolving field with crucial applications in 3D reconstruction and understanding. This task presents considerable challenges, arising from the intrinsic data sparsity incomplete nature of input points generated by LiDAR. paper proposes generative semantic method based on discrete denoising diffusion probabilistic model to tackle these issues. In phase, weighted K-nearest neighbor uniform transition kernel introduced...
Hyperspectral remote sensing has been an important technical means to obtain more refined information and provide rich, accurate, reasonable data for the quantitative analysis delicacy management of a "smart city". To better understand use hyperspectral help construction digital city, study feature characteristics images is introduced in this paper. Then how collect urban ground objects through unmanned aerial vehicle (UAV) imager was described, which greatly improves efficiency acquisition....
In the augmented reality system, position and direction of user's point view line sight in real scene is acquired real-time. The information will determine exact virtual object scene. At same time, various coordinate systems are established according to sight. So registration tracking technology very important. paper proposes an accurate, stable, effective algorithm. method adopts ORB (oriented FAST rotated BRIEF) features matching combined with RANSAC (random sample consensus) obtain...
Some problems still exist in health management and application such as insufficient data, limited technology, lack of professional evaluation methods by physicians with medical theory. In this study, an intelligent method is based on analysis physiological big data collected wearable smartwatches. Firstly, pulse, heart rate, blood oxygen were continuously from individuals wearing smartwatches, the was digitally transmitted. Secondly, transmitted sent to a platform Narrow Band Internet...
The last two decades have witnessed a massive increase in research pertaining to the wide range of use artificial diamond films, particular, chemical vapor deposition (CVD) coatings. Exclusive attention has been given these coatings as result their immense biocompatibility, biofunctionality and non-toxic properties, thereby justifying preference over other conventional long-used biomedical materials. This review study details chemistry, growth mechanism, forms determinants CVD sole purpose...
Scientific visualization is important in modern technological activities and engineering exploration. Due to the dark high-pressure characteristics of deep sea, it difficult visualize entire operation deep-sea equipment. Thus, great necessity use virtual simulation technology help people understand process some exploration equipment on sea floor. In this paper, science, art, new media are combined through artistic rendering, visual processing, reality (VR) holography, which makes latest...
The rapid and accurate identification of various road traffic signs is an important research topic in automotive vision systems. Specifically, the correct urgent problem requiring effective solutions to facilitate automatic driving. This paper proposes a new approach, PP-LCNet-P2-CT, for detection recognition urban system using improved YOLOv5 deep learning model. main improvement PP-LCNet-P2-CT model includes following: (1) Replacing backbone network with lightweight PP-LCNet improve...
The rapid growth in digital content has necessitated advanced retrieval systems capable of efficiently managing and searching through multimodal data. This paper introduces the Cascading Fusion Cross-attention Model for Visual Textual Retrieval, a novel approach designed to enhance accuracy retrieving textual descriptions corresponding visual data vice versa. model employs unique cascading fusion strategy that integrates multiple levels feature representations from both modalities, ensuring...
In the smart city construction and development process, network security is a vital link that must be addressed. As critical technology to ensure of cities, intrusion detection systems play multiple roles, such as real-time monitoring, threat identification, event response. By deploying optimizing efficient mechanisms, cities can effectively resist various attacks safety stability urban operations. Therefore, this paper proposes PNN feature selection model based on binary particle swarm...
It is a hotspot to retrieve the desired information or find specific target inside mass traffic video information.To achieve content, it necessary build up bridge between low-level visual and high-level semantics.In this paper, firstly we propose framework of intelligent semantic description.And then description based on standard MPEG-7 brought out.The can describe multi-layer from underlying features features, make gap ultimately facilitate users information.Finally, accuracy rapidity...