A5100384088- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Advanced Graph Neural Networks
- Online Learning and Analytics
- Topic Modeling
- Traditional Chinese Medicine Studies
- Advanced Image Fusion Techniques
- Intelligent Tutoring Systems and Adaptive Learning
- Image and Object Detection Techniques
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Artificial Intelligence in Healthcare
- Machine Learning in Healthcare
- Complex Network Analysis Techniques
- Image and Signal Denoising Methods
- Image Processing and 3D Reconstruction
- Digital Media and Visual Art
- Artificial Intelligence in Education
- Fire Detection and Safety Systems
- Imbalanced Data Classification Techniques
- Ultrasound Imaging and Elastography
- Video Coding and Compression Technologies
North China University of Water Resources and Electric Power
2023-2025
Hohai University
2024
Beijing Microelectronics Technology Institute
2024
Tianjin University of Technology and Education
2023-2024
Shangqiu Normal University
2023
East China Normal University
2023
Northeastern University
2022-2023
Minzu University of China
2023
Peking University
2020-2021
AVIC Optronics (China)
2021
Object detection on the drone faces a great diversity of challenges such as small object inference, background clutter and wide viewpoint. In contrast to traditional problem in computer vision, bird-like angle can not be transplanted directly from common-in-use methods due special texture sky's view. However, lack comprehensive data set, number algorithms that focus using captured by drones is limited. So VisDrone team gathered massive set organized Vision Meets Drones: A Challenge...
Small object is one of the primary challenges in field detection, which notably pronounced to detection images from Unmanned Aerial Vehicles (UAV). Existing detectors based on deep-learning methods usually apply feature extraction networks with a large down-sampling factor obtain higher-level features. However, such big stride tends make information small objects become little point or even vanish low-resolution maps due limitation pixels. Therefore, novel structure called Multi-branch...
Recommender systems are widely used in e-commerce, movies, music, social media, and other fields because of their personalized recommendation functions. The algorithm is to capture user preferences, item characteristics, the items that users interested recommended users. Matrix factorization collaborative filtering algorithms its simplicity efficiency. However, simple dot-product method cannot establish a nonlinear relationship between latent features or make full use information. model...
In order to make the intelligent vehicle run safely and improve applicability of local path planning algorithm in vehicle, paper presents a method for vehicles. Firstly, establish model Ackerman steering. addition, minimum turning radius constraint is added speed screening mechanism based on traditional dynamic window approach. Then, avoid excessive changes driving vehicles, curvature retention evaluation index function. The simulation results show that improved meets requirements obstacle...
After years of development, the traditional decision tree algorithm implementations represented by LightGBM has been very mature and widely used in various classification problems. However, when applying to detecting fraud health insurance data, we find that performance is not ideal due large imbalance between number examples normal examples. To solve this problem, propose a simple effective LightGBM-based hard example mining (LHEM) for fraud. Our motivation detect small dataset. Selecting...
Uterine-cancer is one of the most common malignant gynecological tumors, a serious threat to women health. The early symptoms uterine cancer: its entire uterus often accompanied by many clusters fine, granular micro calcifications and individual grains appearance. Study on extracting has important significance in diagnosis clinical medicine. Due size shape variability calcifications, low contrast back ground ultrasonic image texture un-homogeneous, so it leads result there exits no general...
Inspired by the tremendous success of self-attention mechanism in natural language processing, Vision Transformer (ViT) creatively applies it to image patch sequences and achieves incredible performance. However, scaled dot-product ViT brings about scale ambiguity structure original feature space. To address this problem, we propose a novel method named Orthogonal (O-ViT), optimize from geometric perspective. O-ViT limits parameters blocks be on norm-keeping orthogonal manifold, which can...
For the purpose of information management on postmark according to date, paper put forward a method date recognition based machine vision, which could meet demands personal collectors. On basis relative theories image processing and pattern recognition, overall process is introduced in from acquisition recognition. Firstly, threshold used generate binary smoothed image. So region numbers be extracted different features. Then regions are connected or broken processed through mathematical...
Knowledge proficiency refers to the extent which students master knowledge and reflects their cognitive status. To accurately assess proficiency, various pedagogical theories have emerged. Bloom’s theory, proposed in 1956 as one of classic theories, follows progression from foundational advanced levels, categorizing cognition into multiple tiers including “knowing”, “understanding”, “application”, thereby constructing a hierarchical structure. This theory is predominantly employed frame...
Introducing Group Equivariant Convolution (GConv) empowers models to explore symmetries hidden in visual data, improving their performance. However, real-world scenarios, objects or scenes often exhibit perturbations of a symmetric system, specifically deviation from architecture, which can be characterized by non-trivial action symmetry group, known as Symmetry-Breaking. Traditional GConv methods are limited the strict operation rules group space, only ensuring features remain strictly...
Patient similarity assessment (PSA) is pivotal to evidence-based and personalized medicine, enabled by analyzing the increasingly available electronic health records (EHRs). However, machine learning approaches for PSA has deal with inherent data deficiencies of EHRs, namely missing values, noise, small sample sizes. In this work, an end-to-end discriminative framework, called SparGE, proposed address these challenges EHR PSA. SparGE measures jointly sparse coding graph embedding. First, we...
Abstract In the field of text classification, most previous work only uses one-hot labels, ignoring correlations between labels. The paper proposes a novel label-enhanced classification model, which utilizes semantic correlation sentences and category labels in Natural Language Processing (NLP) task to integrate label information. We measure similarity instance with among test proposed model on tasks two levels: (document level) sentiment analysis (sentence level). Experimental results show...
This paper proposes a novel hybrid method to reduce speckle noise in ultrasonography.This applies the total variation denoising algorithm output image of recently reported anisotropic diffusion filter.Performance proposed is illustrated using simulated and clinical images.Experimental results indicate outperforms existing despeckling schemes terms both reduction edge preservation.