A5101924913- Infrared Target Detection Methodologies
- Advanced Image and Video Retrieval Techniques
- 3D Surveying and Cultural Heritage
- Robotics and Sensor-Based Localization
- Advanced Neural Network Applications
- 3D Shape Modeling and Analysis
- Thermography and Photoacoustic Techniques
- Remote Sensing and LiDAR Applications
- Advanced Semiconductor Detectors and Materials
- Visual Attention and Saliency Detection
- CCD and CMOS Imaging Sensors
- Autonomous Vehicle Technology and Safety
- Advanced Measurement and Detection Methods
- Air Quality Monitoring and Forecasting
- Video Surveillance and Tracking Methods
- Advanced Image Fusion Techniques
- Advanced Sensor and Control Systems
- Traffic control and management
- Olfactory and Sensory Function Studies
- Human Pose and Action Recognition
- Advanced SAR Imaging Techniques
- Indoor and Outdoor Localization Technologies
- Advanced Neuroimaging Techniques and Applications
- Image Enhancement Techniques
- Infrared Thermography in Medicine
Nanjing University of Science and Technology
2015-2024
Xi'an University of Architecture and Technology
2020-2021
National University of Defense Technology
2021
Wuhan University
2007-2015
Xi'an Jiaotong University
2009
Beihang University
2009
Infrared small target detection (ISTD) has a wide range of applications in early warning, rescue, and guidance. However, CNN based deep learning methods are not effective at segmenting infrared (IRST) that it lack clear contour texture features, transformer also struggle to achieve significant results due the absence convolution induction bias. To address these issues, we propose new model called attention with bilinear correlation (ABC), which is on architecture includes linear fusion...
In the point-cloud-based place recognition area, existing hybrid architectures combining both convolutional networks and transformers have shown promising performance. They mainly apply voxel-wise transformer after sparse convolution (SPConv). However, they can induce information loss by voxelization further result in propagation to transformer, significantly degrading performance of network, especially outdoor scenes with complex geometric structures multiple small objects. To address this...
Patch image model has recently shown significant superiority in the detection of infrared small and dim targets. In this paper, we incorporate more useful local global information into sophisticated patch-image called reweighted patch-tensor model, for its efficiency flexibility. Local signal-clutter-ratio analysis is employed to enhance targets avoid being overwhelmed by strong background edges. meantime, nuclear norm minimization applied globally measure low-rank property a couple matrixes...
Transformer has recently become widely adopted in point cloud registration. Nevertheless, is unsuitable for handling dense clouds due to resource constraints and the sheer volume of data. We propose a method directly regressing rigid relative transformation pairs. Specifically, we divide into blocks according down-sampled superpoints. During training, randomly select with varying overlap ratios, during testing, introduce overlap-aware Rotation-Invariant Geometric Cross-Encoder...
Tracing software entity dependencies is a difficult and time-consuming task, the incomplete changes on systems are prone to induce bugs. Mining frequent itemset widely used find co-changed entities, with which can be detected. In this paper, we present an improved method predict entities in context of evolution. order extract change transactions precisely, customized extraction algorithm for transaction fuzzy matching strategy proposed our approach, then Apriori reduced mining patterns...
Point-cloud-based place recognition is a key component for outdoor large-scale Simultaneous Localization And Mapping (SLAM) in re-localization. However, most methods have limited generalization ability unseen environments. To address this issue, Hybrid Voxel- and Point- wise network, named HVP-Net, proposed. This network utilizes sparse convolutions to learn the local detail of voxel- features proposed lightweight grouped efficient attention mechanisms capture global representations point-...
Video Snapshot Compressive Imaging (SCI) uses a low-speed 2D camera to capture high-speed scenes as snapshot compressed measurements, followed by reconstruction algorithm retrieve the video frames. The fast evolving mobile devices and existing high-performance SCI algorithms motivate us develop methods for real-world applications. Yet, it is still challenging deploy previous on due complex inference process, let alone real-time reconstruction. To best of our knowledge, there no model...
Recent works have shown that neural radiance fields (NeRFs) on top of parametric models reached SOTA quality to build photorealistic head avatars from a monocular video. However, one major limitation the NeRF-based is slow rendering speed due dense point sampling NeRF, preventing them broader utility resource-constrained devices. We introduce LightAvatar, first avatar model based light (NeLFs). LightAvatar renders an image 3DMM parameters and camera pose via single network forward pass,...
Infrared small target detection faces the inherent challenge of precisely localizing dim targets amidst complex background clutter. Traditional approaches struggle to balance precision and false alarm rates. To break this dilemma, we propose SeRankDet, a deep network that achieves high accuracy beyond conventional hit-miss trade-off, by following ``Pick Bunch'' principle. At its core lies our Selective Rank-Aware Attention (SeRank) module, employing non-linear Top-K selection process...
Water hazard detection is an important yet challenging task in autonomous driving as the complex underwater geography brings many hidden risks, e.g. puddles, which could make self-driving cars unsafe. Fully convolutional networks (FCN) have achieved remarkable performance on image segmentation tasks, but water problems are always hard to deal with due reflection characteristic of water. In this paper, we use Conditional Generative Adversarial Networks (cGAN) detection. It has been proved...
Efficient filtration provides a healthy and comfortable residential environment. Based on the performance testing for filter materials commonly used in China, mass balance equation of controlling indoor PM2.5 concentration was set up, selection criterion discussed. An optimal section suggested by taking into account economic cost, price, air quality.