A5101671931- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Video Analysis and Summarization
- Human Pose and Action Recognition
- Advanced Image Fusion Techniques
- Advanced Vision and Imaging
- Online Learning and Analytics
- Image and Signal Denoising Methods
- Video Surveillance and Tracking Methods
- Education and Learning Interventions
- Image Enhancement Techniques
- Teaching and Learning Programming
- Virtual Reality Applications and Impacts
- Advanced Image Processing Techniques
- Medical Image Segmentation Techniques
- Intelligent Tutoring Systems and Adaptive Learning
- Crop Yield and Soil Fertility
- Infrared Target Detection Methodologies
- Remote Sensing and Land Use
- Digital Holography and Microscopy
- 3D Shape Modeling and Analysis
- Software Reliability and Analysis Research
- Image Retrieval and Classification Techniques
- Computer Graphics and Visualization Techniques
- Rice Cultivation and Yield Improvement
Zhejiang Gongshang University
2020-2024
East China Normal University
2024
Heilongjiang Bayi Agricultural University
2024
Wenzhou University
2023
Shenyang Aerospace University
2020
Zhejiang Normal University
2016
Out-of-focus blur is a common image degradation phenomenon that occurs in case of lens defocusing. The out-of-focus kernel usually modeled as Gaussian function or uniform disk previous work. In this paper, we propose it can be more accurately depicted using the generalized (GG) function. This motivated by theoretical analysis and practical observation real kernels. We show kernels are specific shapes, GG further simplified to single-parameter model. estimate parameter from patches containing...
Current methods for text-to-video retrieval (T2VR) are trained and tested on video-captioning oriented datasets such as MSVD, MSR-VTT VATEX. A key property of these is that videos assumed to be temporally pre-trimmed with short duration, whilst the provided captions well describe gist video content. Consequently, a given paired caption, supposed fully relevant caption. In reality, however, queries not known priori, clips may contain sufficient content meet query. This suggests gap between...
This paper targets unsupervised skeleton-based action representation learning and proposes a new Hierarchical Contrast (HiCo) framework. Different from the existing contrastive-based solutions that typically represent an input skeleton sequence into instance-level features perform contrast holistically, our proposed HiCo represents multiple-level performs in hierarchical manner. Specifically, given human sequence, we it multiple feature vectors of different granularities both temporal...
Virtual-reality 3D modeling helps primary school students to develop creative thinking and problem-solving skills. Through hands-on practice, can understand abstract concepts more intuitively, realize the combination of theory practice. However, in conventional virtual teaching, often lack immersive experience, method may not be line with cognitive way students, which, turn, causes high load. Immersive reality (IVR) environments provide intuitive interactions, which help promote students’...
Programming education is gaining attention at the K-12 level. In digital era, computational thinking seen as a key skill. Students in programming debugging process can not only fix code errors but also exercise and cultivate thinking. However, learners level lack confidence due to of foundational knowledge difficulty obtaining effective feedback environment. The emergence large language models (LLMs) provides new pathway for novice training. This study applied advantages these debugging,...
Cross-spectral image guided denoising has shown its great potential in recovering clean images with rich details, such as using the near-infrared to guide process of visible one. To obtain pairs, a feasible and economical way is employ stereo system, which widely used on mobile devices. Current works attempt generate an aligned guidance handle disparity between two images. However, due occlusion, spectral differences noise degradation, generally exists ghosting artifacts, leading...
The development of combinatorial adjuvants is a promising strategy to boost vaccination efficiency. Accumulating evidence indicates that manganese exerts strong immunocompetence and will become an enormous potential adjuvant. Here, we described novel combination Mn
We propose a novel unsupervised cross-modal homography estimation framework based on intra-modal Self-supervised learning, Correlation, and consistent feature map Projection, namely SCPNet. The concept of self-supervised learning is first presented to facilitate the estimation. correlation-based network projection are combined form learnable architecture SCPNet, boosting framework. SCPNet achieve effective satellite-map image pair dataset, GoogleMap, under [-32,+32] offset 128x128 image,...
Abstract Achieving high‐performance in multi‐object tracking algorithms heavily relies on modelling spatial‐temporal relationships during the data association stage. Mainstream approaches encompass rule‐based and deep learning‐based methods for relationship modelling. While former physical motion laws, offering wider applicability but yielding suboptimal results complex object movements, latter, though achieving high‐performance, lacks interpretability involves module designs. This work aims...
Multispectral and multimodal images are of important usage in the fieldof multi-source visual information fusion. Due to alternation or movement image devices, acquired multispectral usually misaligned, hence registration is pre-requisite. Different from common images, a challenging problem due nonlinear variation intensity gradient. To cope with this challenge, we propose phase congruency network (PCNet) enhance structure similarity images. The can then be aligned using similarity-enhanced...
We study the dynamical properties of quantum Rabi model within a systematic expansion method. Based on observation that parity symmetry is kept during evolution states, we decompose initial state and time-dependent one into part positive negative expanded by superposition coherent states. The evolutions for corresponding are obtained, where coefficients in equations known from recurrence relation derived.
This paper makes a perfect and reasonable explanation solution to the highly researchable problems under melting of silica, uses methods image matching, moving state target tracking, gray binary construct multiple linear regression. Equations, feature matching models, background difference etc., are comprehensively solved using software such as MATLAB, it is found that silica particle coordinates motion trajectories, generalized radius area can represent changes in process conclusion rate...
There are a variety of factors on stage environment such as dramatically changing lights, mutation and similar appearance performers that make computer vision based performer detection task challenging. Relighting technique is adopted to handle the lights problem can greatly relieve negative effects caused by results. Color transfer one commonly used relighting techniques in task. However, existing color methods will bring artifacts like structural defects, distortions chromatic noises...
This paper targets unsupervised skeleton-based action representation learning and proposes a new Hierarchical Contrast (HiCo) framework. Different from the existing contrastive-based solutions that typically represent an input skeleton sequence into instance-level features perform contrast holistically, our proposed HiCo represents multiple-level performs in hierarchical manner. Specifically, given human sequence, we it multiple feature vectors of different granularities both temporal...