A5100785431- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- Human Pose and Action Recognition
- Advanced Neural Network Applications
- Image Enhancement Techniques
- Remote-Sensing Image Classification
- Video Surveillance and Tracking Methods
- Anomaly Detection Techniques and Applications
- Magnetic Bearings and Levitation Dynamics
- Image and Signal Denoising Methods
- CCD and CMOS Imaging Sensors
- 3D Shape Modeling and Analysis
- Image Retrieval and Classification Techniques
- Computational Geometry and Mesh Generation
- Handwritten Text Recognition Techniques
- Advanced Image Fusion Techniques
- Digital Image Processing Techniques
- Data Management and Algorithms
- Advanced Image Processing Techniques
- Robotics and Sensor-Based Localization
- Bone Tumor Diagnosis and Treatments
- Dental Radiography and Imaging
- Hand Gesture Recognition Systems
- Blockchain Technology Applications and Security
- Visual Attention and Saliency Detection
Xidian University
2024-2025
Children's Hospital of Fudan University
2025
Fudan University
2022-2024
Heilongjiang University
2024
Zhengzhou University of Light Industry
2023-2024
Guangzhou Huali College
2024
University of Electronic Science and Technology of China
2016-2023
Beijing Normal University - Hong Kong Baptist University United International College
2022-2023
Liaoning Normal University
2021-2022
Chinese Academy of Sciences
2022
Video super-resolution has recently become one of the most important mobile-related problems due to rise video communication and streaming services. While many solutions have been proposed for this task, majority them are too computationally expensive run on portable devices with limited hardware resources. To address problem, we introduce first Mobile AI challenge, where target is develop an end-to-end deep learning-based that can achieve a real-time performance mobile GPUs. The...
Weakly supervised object detection (WSOD) in remote sensing images is used to detect high-value objects by utilizing image-level labels. However, the current models still have two problems. Firstly, misclassification of neighboring instances easily occurred because one-hot label assigned all seed and their instances. Secondly, supervisory information each instance classifier refinement (ICR) branch generated from predicted class score upper ICR rather than real label, thus prediction mistake...
Weakly supervised object detection (WSOD) has a great practical value in remote sensing image (RSI) interpretation because the instance-level annotations are not required. The multiple instance learning based methods mainstream, and two problems should be addressed. First of all, majority usually detect discriminative parts rather than whole object. Secondly, quantity easy instances is much greater that hard instances, which restricts improvement WSOD methods. To address first problem,...
Although viral infections are one of the common clinical manifestations in patients with inborn errors immunity (IEIs), little is known about epidemiology, susceptibility genes, and status IEIs. The demographic information, diagnoses, laboratory findings 931 IEI who underwent testing from January 2016 to December 2022 were collected analyzed. In total, 47.15% (439/931) tested positive for at least virus during hospitalization. There a total 640 study period, mainly EBV 131 (20.47%), HRV...
Weakly supervised object detection (WSOD) in remote sensing images (RSIs) aims to achieve high-value classification and localization using only image-level labels, it has a wide range of applications. However, existing popular WSOD models still encounter two challenges. First, these typically select the highest-scoring proposal as seed instance while ignoring lower-scoring ones, resulting some less-obvious objects being missed. Second, current fail ensure consistency between regression,...
State-of-the-art temporal action detectors to date are based on two-stream input including RGB frames and optical flow. Although combining flow boosts performance significantly, is a hand-designed representation which not only requires heavy computation, but also makes it methodologically unsatisfactory that methods often learned end-to-end jointly with the In this paper, we argue dispensable in high-accuracy detection image level data augmentation (ILDA) key solution avoid degradation when...
How to effectively extract spectral and spatial information apply it hyperspectral image classification (HSIC) has been a hot research topic. In recent years, the transformer-based HSIC models have attracted much interest due their advantages in long-distance modeling of features images (HSIs). However, method suffers from high computational complexity, especially tasks that require processing large amounts data. addition, variability inherent HSIs limits performance improvement HSIC. To...
This paper presents a novel solution of vehicle occlusion and 3D measurement for traffic monitoring by data fusion from multiple stationary cameras. Comparing with single camera based conventional methods in monitoring, our approach fuses video different viewpoints into common probability map (PFM) extracts targets. The proposed PFM concept is efficient to handle fuse order estimate the appearance, which verified be more reliable than real outdoor experiments. An AMF shadowing modeling...
Abstract Human taste perception is associated with the papillae on tongue as they contain a large proportion of chemoreceptors for basic tastes and other chemosensation. Especially density fungiform (FP) considered an index responsiveness to oral chemosensory stimuli. The standard procedure FP counting involves visual identification manual specific parts by trained operators. This tedious task automated image analysis methods are desirable. In this paper machine learning processing method...
This paper presents a general frame to integrate segmentation and recognition gives novel method identify lingual attribute of mixed Chinese/English characters. The outstanding performance this is as follows. First, text- line rather than character segment regarded process unit. Second, multi-feature adopted based on multi-phase segmentation. Third, two types feedbacks, including from feature statistic within text-line, are throughout the whole recognition. Fourth, it adaptive quality genre...
Notice of Violation IEEE Publication Principles<br><br>"A Color Image Segmentation Method Based on Automatic Seeded Region Growing,"<br>by Weiwei Li, Huixian Huang, Dongbo Zhang, Hongzhong Tang and Chenhao Wang,<br>in the Proceedings 2007 International Conference Automation Logistics, pp. 1925-1927<br><br>After careful considered review content authorship this paper by a duly constituted expert committee, has been found to be in violation IEEE's Principles.<br><br>This contains portions...
Today, Internet service deployment is typically implemented with server replication at multiple locations. Domain name system (DNS), which translates human-readable domain names into network-routable IP addresses, used for distributing users to different replicas. However, DNS relies on several network-based queries and the delay connection setup process between client replica. In this article, we propose Artemis, a practical low-latency naming routing that supports optimal (replica)...
This paper presents a high-speed video transfer scheme and real-time infrared spots detection algorithm designed for field programmable gate array (FPGA) implementation. Rather than IEEE 1394a, two 1394b interfaces are alternatively used to ensure high-resolution image in real time. In order execute fast detection, parallel that processes four pixels per clock cycle is proposed. It detects single pass over frame its implementation only composed of combinatorial logic registers. Furthermore,...
Graph Contrastive Learning (GCL) has attracted significant research attention due to its self-supervised ability learn robust node representations. Unfortunately, most methods primarily focus on homophilic graphs, rendering them less effective for heterophilic graphs. In addition, the complexity of interactions in graphs poses considerable challenges augmentation schemes, coding architectures, and contrastive designs traditional GCL. this work, we propose HeterGCL, a novel graph learning...
We study a variation of facility location problems (FLPs) that aims to improve the accessibility agents within context mechanism design without money. In such variation, have preferences on ideal locations real line, and facility's is fixed in advance where (re)locating not possible due various constraints (e.g., limited space construction costs). To facilities, existing literature FLPs has proposed structurally modify line by adding new interval) or provide shuttle services between two...