A5101866439- Advanced Image Processing Techniques
- Image and Signal Denoising Methods
- Face and Expression Recognition
- Image Enhancement Techniques
- Remote-Sensing Image Classification
- Smart Grid Energy Management
- Face recognition and analysis
- Biometric Identification and Security
- IoT-based Smart Home Systems
- Advanced Optical Sensing Technologies
- Industrial Automation and Control Systems
- Energy Load and Power Forecasting
- Sleep and Work-Related Fatigue
- Advanced Image and Video Retrieval Techniques
- High voltage insulation and dielectric phenomena
- Infrared Target Detection Methodologies
- Advanced Steganography and Watermarking Techniques
- Data-Driven Disease Surveillance
- Image Processing Techniques and Applications
- Geophysical Methods and Applications
- Advanced Image Fusion Techniques
- Visual Attention and Saliency Detection
- Vehicle Dynamics and Control Systems
- Advanced Clustering Algorithms Research
- Advanced Fiber Optic Sensors
Wuhan University
2019-2025
Taiyuan University of Technology
2023-2025
Shanghai Jiao Tong University
2022
Zhejiang Gongshang University
2021
University of Michigan–Dearborn
2015
Northwest University
1991
The loads that have several working states cannot be accurately distinguished by the conventional Non-Intrusive Load Monitoring (NILM) methods. This paper proposed an improved NILM method based on Resnet18 Convolutional Neural Network (CNN) and Support Vector Machine (SVM) algorithm to address misidentification of multi-state appliances. V-I trajectories are at first classified with Resnet18. Then, load features low redundancy is obtained through Max-Relevance Min-Redundancy (mRMR) feature...
Accurate hyperspectral image (HSI) interpretation is critical for providing valuable insights into various earth observation-related applications such as urban planning, precision agriculture, and environmental monitoring. However, existing HSI processing methods are predominantly task-specific scene-dependent, which severely limits their ability to transfer knowledge across tasks scenes, thereby reducing the practicality in real-world applications. To address these challenges, we present...
Image restoration aims to reconstruct degraded images, e.g., denoising or deblurring. Existing works focus on designing task-specific methods and there are inadequate attempts at universal methods. However, simply unifying multiple tasks into one architecture suffers from uncontrollable undesired predictions. To address those issues, we explore prompt learning in architectures for image tasks. In this paper, present Degradation-aware Visual Prompts, which encode various types of degradation,...
The paper proposed a model using real time driving front video recording to detect driver drowsiness. recordings were fed into the TRW's simulator obtain lane-related signals. Time domain features and frequency extracted from signals characterize difference of alert state drowsiness state. Both support vector machine neural network used Experimental results on word illustrated that method can with good accuracy. It also show TRW generate reliable lane related if high quality sequences are provided.
In order to get raw images of high quality for downstream Image Signal Process (ISP), in this paper we present an Efficient Locally Multiplicative Transformer called ELMformer image restoration. contains two core designs especially whose primitive attribute is single-channel. The first design a Bi-directional Fusion Projection (BFP) module, where consider both the color characteristics and spatial structure second one that propose Self-Attention (L-MSA) scheme effectively deliver information...
Non-negative Matrix Factorization (NMF) and spectral clustering have been proved to be efficient effective for data tasks applied various real-world scenes. However, there are still some drawbacks in traditional methods: (1) most existing algorithms only consider high-dimensional directly while neglect the intrinsic structure low-dimensional subspace; (2) pseudo-information got optimization process is not relevant manifold regularization methods. In this paper, a novel unsupervised matrix...
Face recognition has achieved a great success in recent years, it is still challenging to recognize those facial images with extreme poses. Traditional methods consider as domain gap problem. Many of them settle by generating fake frontal faces from ones, whereas they are tough maintain the identity information high computational consumption and uncontrolled disturbances. Our experimental analysis shows dramatic precision drop Meanwhile, poses just exist minor visual differences after small...
The limitations of task-specific and general image restoration methods for specific degradation have prompted the development all-in-one techniques. However, diversity patterns among multiple degradation, along with significant uncertainties in mapping between degraded images different severities their corresponding undistorted versions, pose challenges to tasks. To address these challenges, we propose Perceive-IR, an restorer designed achieve fine-grained quality control that enables...
Collecting real-world mobility data is challenging. It often fraught with privacy concerns, logistical difficulties, and inherent biases. Moreover, accurately annotating anomalies in large-scale nearly impossible, as it demands meticulous effort to distinguish subtle complex patterns. These challenges significantly impede progress geospatial anomaly detection research by restricting access reliable complicating the rigorous evaluation, comparison, benchmarking of methodologies. To address...
Few-shot learning aims to recognize novel concepts by leveraging prior knowledge learned from a few samples. However, for visually intensive tasks such as few-shot semantic segmentation, pixel-level annotations are time-consuming and costly. Therefore, in this paper, we utilize the more challenging image-level propose an adaptive frequency-aware network (AFANet) weakly-supervised segmentation (WFSS). Specifically, first cross-granularity module (CFM) that decouples RGB images into...
Renovating the memories in old photos is an intriguing research topic computer vision fields. These legacy images often suffer from severe and commingled degradations such as cracks, noise, color-fading, while lack of large-scale paired photo datasets makes this restoration task very challenging. In work, we present a novel reference-based end-to-end learning framework that can jointly repair colorize degraded pictures. Specifically, proposed consists three modules: sub-network for...
Aiming at the fact that traditional non-intrusive load monitoring and identification methods cannot accurately identify household appliances containing multiple operating states, this paper proposes an method for loads based on multi-state feature information fusion. Firstly, mRMR selection algorithm is used to select a set of optimal vectors from each state as input, then SVM identification, which achieves classification effect quickly finely identifying confusable home appliances. Finally,...
Acoustic signals produced by an operated transformer are the important parameters for assessment of noise level substation and condition monitoring transformer. When acoustic transmit in air, some interference components inevitably contained measured microphone sensors with features diversity a certain degree uncertainty. To obtain real resulting from transformer, adversarial convolutional de-noising auto-coder (CDAE) network is proposed trained to carefully depress end-to-end manner. The...
Abstract ChemInform is a weekly Abstracting Service, delivering concise information at glance that was extracted from about 100 leading journals. To access of an article which published elsewhere, please select “Full Text” option. The original trackable via the “References”