A5067715727- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Image Processing Techniques and Applications
- Advanced SAR Imaging Techniques
- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Remote-Sensing Image Classification
- Image Enhancement Techniques
- Visual Attention and Saliency Detection
- Advanced Vision and Imaging
- Text and Document Classification Technologies
- Remote Sensing and Land Use
- Software Engineering Research
- Medical Image Segmentation Techniques
- Anomaly Detection Techniques and Applications
- Advanced Image Fusion Techniques
- Domain Adaptation and Few-Shot Learning
- Geophysical Methods and Applications
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Optical Systems and Laser Technology
- Advanced Optimization Algorithms Research
- Distributed and Parallel Computing Systems
- Advanced Computational Techniques and Applications
- Advanced Sensor and Control Systems
- Geochemistry and Geologic Mapping
Chinese Academy of Sciences
2010-2025
Zhejiang Normal University
2025
Tianjin University
2017-2025
Wuyi University
2025
Shanghai Institute of Optics and Fine Mechanics
2025
Beijing Institute of Technology
2023-2024
China University of Petroleum, Beijing
2024
Institute of Automation
2017-2024
Wuyi University
2020-2024
Ministry of Transport
2024
In this paper, a parallel network based on hand detection and body pose estimation is proposed to detect distinguish human's right left hands. The employed for human–robot interaction (HRI) gestures. This method fully uses feature information in the human structure. One channel ResNet-Inception-Single Shot MultiBox Detector extract detection. other estimates first then positions of hands using forward kinematic tree skeleton Thereafter, results two channels are fused. fusion module,...
Accurate segmentation of the prostate is a key step in external beam radiation therapy treatments. In this paper, we tackle challenging task CT images by two-stage network with 1) first stage to fast localize, and 2) second accurately segment prostate. To precisely stage, formulate into multi-task learning framework, which includes main prostate, an auxiliary delineate boundary. Here, applied provide additional guidance unclear boundary images. Besides, conventional deep networks typically...
Multiimage super-resolution (MISR), as one of the most promising directions in remote sensing, has become a needy technique satellite market. A sequence images collected by satellites often plenty views and long time span, so integrating multiple low-resolution into high-resolution image with details emerges challenging problem. However, MISR methods based on deep learning cannot make full use images. Their fusion modules are incapable adapting to an weak temporal correlations well. To cope...
Ship tracking technology is crucial for emergency rescue in the event of a disaster. Quickly identifying position and status vessels vital teams to be able deploy efficiently disaster areas. When responding emergencies or natural disasters, ship plays critical role supporting operations resource allocation, improving overall resilience maritime transportation system. However, research on multi-object (MOT) algorithms has primarily focused optical image datasets. In contrast, data from...
Athlete engagement is influenced by several factors, including cohesion, passion and mental toughness. Machine learning methods are frequently employed to construct predictive models as a result of their high efficiency. In order comprehend the effects toughness on athlete engagement, this study utilizes relevant machine prediction model, so find intrinsic connection between them. The construction comparison algorithms investigated evaluate level in determine optimal model. results show that...
With the heated trend of augmented reality (AR) and popularity smart head-mounted devices, development natural human device interaction is important, especially hand gesture based interaction. This paper presents a solution for point in egocentric vision its application. Firstly, dataset named EgoFinger established focusing on pointing vision. We discuss collection detail as well comprehensive analysis this dataset, including background foreground color distribution, occurrence likelihood,...
During the real-aperture-scanning imaging process, terahertz (THz) images are often plagued with problem of low spatial resolution. Therefore, an accommodative super-resolution framework for THz is proposed. Specifically, 3D degradation model system firstly proposed by incorporating focused beam distribution, which determines relationship between range and corresponding image restoration level. Secondly, adjustable CNN introduced to cope this dependent problem. By simply tuning interpolation...
Ship detection from synthetic aperture radar (SAR) images is inherently subject to the special imaging mechanism of SAR. In recent years, deep-learning-based techniques for detecting objects optical have rapidly advanced and promoted development SAR image technology. However, strong speckle noise in degrades low-level feature learning shallow layers, hindering higher level semantic features object detection. view problems encountered direct end-to-end close relationship between auxiliary...
Abstract Laccase is capable of catalyzing a vast array reactions, but its low redox potential limits applications. The use photocatalytic materials offers solution to this problem by converting absorbed visible light into electrons facilitate enzyme catalysis. Herein, MIL‐53(Fe) and NH 2 ‐MIL‐53(Fe) serve as both absorbers immobilization carriers, laccase employed for solar‐driven chemical conversion. Electron spin resonance spectroscopy results confirm that irradiation causes rapid transfer...
High-resolution range profile (HRRP) is increasingly employed in radar target recognition under intricate ground scenarios. Such scenarios demand recognizing the specific type of a from wide categories, task known as fine-grained (FGTR), which involves numerous and potentially unbalanced categories. To tackle this, we propose joint semantic-data guided hierarchical classification (SDHC) framework. It consists set local classifiers organized tree hierarchy based on relationship. allows...
With dramatically increasing of the spatial resolution satellite imaging sensors, object-based image analysis (OBIA) has been gaining prominence in remote sensing applications. Multiscale segmentation is a prerequisite step that splits an into hierarchical homogeneous segmented objects for OBIA. However, scale selection remains challenge multiscale segmentation. In this study, we presented adaptive approach defining and estimating optimal process. Central to our method combined use features...
This paper provides efficient and robust algorithms for real-time face detection recognition in complex backgrounds. The are implemented using a series of signal processing methods including Ada Boost, cascade classifier, Local Binary Pattern (LBP), Haar-like feature, facial image pre-processing Principal Component Analysis (PCA). Boost algorithm is classifier to train the eye detectors with accuracy. LBP descriptor utilized extract features fast detection. reduces false rate. detected then...
The skeleton based gesture recognition is gaining more popularity due to its wide possible applications. key issues are how extract discriminative features and design the classification model. In this paper, we first leverage a robust feature descriptor, path signature (PS), propose three PS explicitly represent spatial temporal motion characteristics, i.e., (S PS), (T PS) S PS). Considering significance of fine hand movements in gesture, an ”attention on hand” (AOH) principle define joint...
We present multiple-image encryption (MIE) based on compressive holography. In the encryption, a holographic technique is employed to record multiple images simultaneously form hologram. The two-dimensional Fourier data of hologram are then compressed by nonuniform sampling, which gives rise encryption. Decryption individual cast into minimization problem. retains sparsity recovered in wavelet basis. Meanwhile, total variation regularization used preserve edges reconstruction. Experiments...
The minerals in the hydrothermal and cold seep system form at different temperatures show responses to laser power varying degrees. Here, we focus on heat-induced by study thermal transformations of chalcopyrite, covellite, pyrite, barite, aragonite based Raman spectroscopy. Chalcopyrite mainly transforms into hematite, covellite chalcocite with increase power. Interestingly, comparing previous study, pyrite can transform marcasite firstly, hematite finally. We also find that...
Context . Weak gravitational lensing is one of the most important probes nature dark matter and energy. In order to extract cosmological information from next-generation weak surveys (e.g., Euclid , Roman LSST, CSST) as much possible, accurate measurements shear are required. Aims There existing algorithms measure on imaging data, which have been successfully applied in previous surveys. meantime, machine learning (ML) has widely recognized various astrophysics applications modeling...
High resolution range profile (HRRP) plays a crucial role in radar target recognition. In real-world applications, variations operational conditions during testing, such as changes depression angles, result unsatisfactory performance for HRRP recognition methods. One way to alleviate this issue is augment training data with samples that embody the testing domain style. Therefore, we propose domain-adaptive generation approach based on two-stage denoising diffusion probability model (DDPM)....
Abstract Autism Spectrum Disorders (ASD) are neurodevelopmental disorders that cause people difficulties in social interaction and communication. Identifying ASD patients based on resting-state functional magnetic resonance imaging (rs-fMRI) data is a promising diagnostic tool, but challenging due to the complex unclear etiology of autism. And it difficult effectively identify with single source (single task). Therefore, address this challenge, we propose novel multi-task learning framework...
Abstract Detecting the various developmental stages of strawberries in their natural environment is crucial for modern agricultural robots. Existing methods focus on fruit detection but ignore stage classification. Moreover, they usually requiring substantial computational resources and are not suitable small low-power embedded platforms. To address this problem, we propose YOLO-VDS, a lightweight model based YOLOv5s optimized We introduce Inverse Residual Bottleneck with 3 Convolutions...