A5101677107- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Remote-Sensing Image Classification
- Image Enhancement Techniques
- Topic Modeling
- Advanced Vision and Imaging
- Sparse and Compressive Sensing Techniques
- Natural Language Processing Techniques
- Image and Signal Denoising Methods
- Advanced Neural Network Applications
- Text and Document Classification Technologies
- Image Processing Techniques and Applications
- Machine Learning and Data Classification
- Remote Sensing and Land Use
- Advanced Graph Neural Networks
- Advanced Image Processing Techniques
- Advanced Optimization Algorithms Research
- Radiation Dose and Imaging
- Emotion and Mood Recognition
- Robotics and Sensor-Based Localization
- Face recognition and analysis
- Industrial Vision Systems and Defect Detection
- Digital Holography and Microscopy
- Gaze Tracking and Assistive Technology
China University of Mining and Technology
2024
Jiangsu University of Science and Technology
2024
Sichuan Agricultural University
2024
Hebei University of Technology
2019-2023
Beijing Institute of Technology
2020-2023
Xidian University
2022
Qufu Normal University
2022
Tianjin University
2013-2020
AT&T (United States)
2013
Arbitrary-oriented objects widely appear in natural scenes, aerial photographs, remote sensing images, etc., and thus arbitrary-oriented object detection has received considerable attention. Many current rotation detectors use plenty of anchors with different orientations to achieve spatial alignment ground truth boxes. Intersection-over-Union (IoU) is then applied sample the positive negative candidates for training. However, we observe that selected cannot always ensure accurate detections...
Imaging through scattering media is a fascinating subject in the computational imaging domain. The methods based on speckle correlation have found tremendous versatility. However, darkroom condition without any stray light required because contrast easily disturbed by ambient light, which can lead to reduction object reconstruction quality. Here, we report plug-and-play (PnP) algorithm restore under non-darkroom environment. Specifically, PnPGAP-FPR method established via generalized...
Background modeling is a critical component for various vision-based applications. Most traditional methods tend to be inefficient when solving large-scale problems. In this paper, we introduce sparse representation into the task of large scale stable background modeling, and reduce video size by exploring its 'discriminative' frames. A cyclic iteration process then proposed extract from discriminative frame set. The two parts combine form our Sparse Outlier Iterative Removal (SOIR)...
Intelligent video surveillance is a vital technique in smart city construction, where detection of objects generally achieved by subtracting estimated background from the raw video. Common wisdom estimation focuses on introducing meaningful structure or discriminative hypothesis to sparsity-based objectives. However, relaxation optimization, which always considered most effective solution, definitely leads information loss. So, this article, as preserve more information, new nonconvex...
In conventional wisdom of video modeling, the background is often treated as primary target and foreground derived using technique subtraction. Based on observation that are two sides same coin, we propose to treat them peer unknown variables formulate a joint estimation problem, called Hierarchical modeling Alternating Optimization (HMAO). The motivation behind our hierarchical extensions models better incorporate priori knowledge about disparity between foreground. For background,...
Deep learning models often use a flat softmax layer to classify samples after feature extraction in visual classification tasks. However, it is hard make single decision of finding the true label from massive classes. In this scenario, hierarchical proved be an effective solution and can utilized replace layer. A key issue construct good structure, which very significant for performance. Several works have been proposed address issue, but they some limitations are almost designed...
Product quantization (PQ) seems to have become the most efficient framework of performing approximate nearest neighbor (ANN) search for high-dimensional data. However, almost all existing PQ-based ANN techniques uniformly allocate precious bit budget each subspace. This is not optimal, because data are often evenly distributed among different subspaces. A better strategy achieve an improved balance between distribution and within Motivated by this observation, we propose develop optimized PQ...
Foreground-background separation of surveillance video, that models static background and extracts moving foreground simultaneously, attracts increasing attentions in building a smart city. Conventional techniques towards this always consider the as primary target tend to adopt low-rank constraint its estimator, which provides finite (equal value rank) alternatives when constructing background. However, practical missions, although general sketch is stable, some details change constantly....
Arbitrary-oriented objects widely appear in natural scenes, aerial photographs, remote sensing images, etc., thus arbitrary-oriented object detection has received considerable attention. Many current rotation detectors use plenty of anchors with different orientations to achieve spatial alignment ground truth boxes, then Intersection-over-Union (IoU) is applied sample the positive and negative candidates for training. However, we observe that selected cannot always ensure accurate detections...
Training noise-robust deep neural networks (DNNs) in label noise scenario is a crucial task. In this paper, we first demonstrates that the DNNs learning with exhibits over-fitting issue on noisy labels because of too confidence its capacity. More significantly, however, it also potentially suffers from under-learning samples clean labels. essentially should pay more attention rather than samples. Inspired by sample-weighting strategy, propose meta-probability weighting (MPW) algorithm which...
Deep learning (DL) is widely applied in the field of hyperspectral image (HSI) classification and has proved to be an extremely promising research technique. However, deployment DL-based HSI algorithms mobile embedded vision applications tends limited by massive parameters, high memory costs, complex networks DL models. In this article, we propose a novel, lightweight, non-deep parallel network (HyperLiteNet) address these issues. Based on development trends hardware devices, proposed...
This paper aims to effectively recognize human faces from images, which is an important problem in the multimedia information process. After analyzing related research works, framework of face recognition system illustrated as first, contains training process and testing Particularly, improved PCA algorithm use feature extraction module. The main innovations this lie that, PCA, we utilize a radial basis function construct kernel matrix by computing distance two different vectors, are...
This paper provides a congestion control mechanism under heterogeneous network and analyze the TCP’s represented research achievements current situation problems environment. According to shortage of TCP network, we design algorithm based on model bandwidth estimation solve problem switching. The results validate effectiveness this simulation analysis, which effectively achieve fair competition resource can improve utilization resources
ABSTRACT Unmanned aerial vehicles (UAVs) are becoming important tools for modern management and scientific research of grassland resources, especially in the dynamic monitoring above‐ground biomass (AGB). However, current studies rely mostly on vertical images to construct models, with little consideration given oblique images. Determination image acquisition height often relies experience intuition, but there is limited comparison models estimating across different types. To address this...