A5100425716- Video Surveillance and Tracking Methods
- Anomaly Detection Techniques and Applications
- Industrial Vision Systems and Defect Detection
- Digital Media and Visual Art
- Advanced Neural Network Applications
- Context-Aware Activity Recognition Systems
- Advanced Data Compression Techniques
- Fire Detection and Safety Systems
- Autonomous Vehicle Technology and Safety
- Energy, Environment, Economic Growth
- Tactile and Sensory Interactions
- Image Enhancement Techniques
- Software Engineering Research
- Air Quality and Health Impacts
- Advanced Image Processing Techniques
- Advanced Computing and Algorithms
- Digital Filter Design and Implementation
- Data Visualization and Analytics
- Video Analysis and Summarization
- Mobile Health and mHealth Applications
- Mobile Crowdsensing and Crowdsourcing
- Multimedia Communication and Technology
- Human Pose and Action Recognition
- Environmental Impact and Sustainability
- Body Composition Measurement Techniques
Guilin University of Technology
2025
Beihang University
2021-2024
Soochow University
2024
University of Shanghai for Science and Technology
2024
Zhengzhou University
2019-2021
Northumbria University
2019
New Jersey Institute of Technology
2010
Tsinghua University
2006
Convolutional Neural Network (CNN) based methods generally take crowd counting as a regression task by outputting densities. They learn the mapping between image contents and density distributions. Though having achieved promising results, these data-driven networks are prone to overestimate or underestimate people counts of regions with different patterns, which degrades whole count accuracy. To overcome this problem, we propose an approach alleviate performance differences in regions....
In this paper, we present a method called density-aware convolutional neural network (DensityCNN) to perform the crowd counting task in various crowded scenes. The key idea of DensityCNN is utilize high-level semantic information provide guidance and constraint when generating density maps. To end, implement by adopting multi-task CNN structure jointly learn density-level classification map estimation. learns multi-channel features that are aware distributions input image. This accomplished...
The lack of discernible vehicle contour features in low-light conditions poses a formidable challenge for nighttime detection under hardware cost constraints. Addressing this issue, an enhanced histogram oriented gradients (HOGs) approach is introduced to extract relevant features. Initially, lights are extracted using combination background illumination removal and saliency model. Subsequently, these integrated with template-based delineate regions containing potential vehicles. In the next...
People in crowd scenes often exhibit the characteristic of imbalanced distribution. On one hand, people size varies largely due to camera perspective. far away from look smaller and are likely occlude each other, whereas near larger relatively sparse. other number also greatly same or different scenes. This article aims develop a novel model that can accurately estimate count given scene with To this end, we have proposed an effective multi-level convolutional neural network (MLCNN)...
Nighttime light remote sensing imagery is limited by its single band and low spatial resolution, hindering ability to accurately capture ground information. To address this, a dual-sampling adjustment method proposed enhance nighttime fusing daytime optical images with imagery, generating high-quality color imagery. The results are as follows: (1) Compared traditional the resolution of fusion improved from 500 m 15 while better retaining features distribution light. (2) Quality evaluations...
The emergent large language/multimodal models facilitate the evolution of mobile agents, especially in task UI automation. However, existing evaluation approaches, which rely on human validation or established datasets to compare agent-predicted actions with predefined ones, are unscalable and unfaithful. To overcome these limitations, this paper presents LlamaTouch, a testbed for on-device agent execution faithful, scalable evaluation. By observing that process only transfers states,...
Human body composition (e.g., total water, fat, muscle) determines health to a large extent. The advent of Android has had significant impact on the way people manage information. Taking advantage this emerging technology realize effective management human information can offer some help and reference for medical care. In paper, we propose an Android-based system. Five major subsystems our system were first designed presented. underlying data process flow was also given in detail. We then...
This paper proposed a novel dual voltage scaling layout architecture and design methodology of low power JPEG2000 codec exploiting scaling. Fabricated in SMIC 0.18/spl mu/m 1P6M standard CMOS technology, this is capable compression/decompression with 1280/spl times/1024 pixel (YUV422 full color) at 20 frames/s employing 100MHz operation frequency. And the consumption 465mW @ 1.8V 100MH. Applied technique, dissipation reduced by 28.5%.
Design method is an important part of design, and it product academic educational research practice. This paper discusses different fields visual design their methodological issues, based on the definition communication WOS a search relevant literature Internet, with aim understanding essential properties activities from stratification perspective clarifying norms disciplines involved in communication. The objectives, content, processing, output two types purpose means are analyzed terms...