A5033524169- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Advanced X-ray and CT Imaging
- Advanced Surface Polishing Techniques
- Adversarial Robustness in Machine Learning
- Robotics and Sensor-Based Localization
- Robotic Locomotion and Control
- Ferroelectric and Piezoelectric Materials
- Advanced Image Fusion Techniques
- Image Enhancement Techniques
- Machine Learning and Data Classification
- Iterative Learning Control Systems
- Composite Material Mechanics
- Thermal properties of materials
- Advanced Measurement and Metrology Techniques
- Diamond and Carbon-based Materials Research
- Image and Object Detection Techniques
- Soft Robotics and Applications
- Ocular Surface and Contact Lens
- Optical measurement and interference techniques
- Seismic Imaging and Inversion Techniques
- Visual Attention and Saliency Detection
- Human Pose and Action Recognition
- Advanced Vision and Imaging
- Electronic and Structural Properties of Oxides
Tianjin University
2025
Beijing Institute of Technology
2024
Northeastern University
2023-2024
China National Petroleum Corporation (China)
2024
Nanjing University of Aeronautics and Astronautics
2002-2022
X-ray prohibited items detection is an essential and highly effective method that has been widely used in inspection scenarios. To satisfy the practical demand for high accuracy security auxiliary diagnosis systems, we propose a geometry-aware detector (GADet) based on classic advanced general object detector, YOLOX. Specifically, first analyze discover relationship between diagonal length category images robust distinctive. Therefore, introduce physical constraint (PDLC), which can utilize...
Summary In this paper, a three-dimensional neural network MultiRes-Unet3D is developed for aiming at the multi-resolution characteristics of faults, and considering small proportion fault points or lines occupied in entire seismic data. Also, weighted cross-entropy loss function used learning process balancing between different items, compared with common crossentropy function, which makes algorithm has more reliable identification ability. We produced few synthetic datasets labels using...
Fixed abrasive lapping and polishing (FALP) is a new machining technology was adopted to manufacture hard brittle materials obtain the high productivity because of fixed abrasive. The preparation process pad (FAP) described. FALP K9 glass, mobile panel glass Si were investigated with 5-10 µm diamond abrasives. effect on material removal rate (MRR) surface quality different studied. results show that in same conditions, highest MRR reaches 4428 nm/min, inferior lowest. And roughness Sa 2.10...
High-frequency and accurate state estimation is crucial for biped robots. This paper presents a tightly-coupled LiDAR-Inertial-Kinematic Odometry (LIKO) robot based on an iterated extended Kalman filter. Beyond estimation, the foot contact position also modeled estimated. allows both velocity updates from kinematic measurement. Additionally, use of measurement results in increased output frequency about 1kHz. ensures temporal continuity estimated makes it practical control purposes We...
With the continuous advancement of computer vision technology and sophisticated projection equipment, method projecting onto special-shaped surfaces has become a current research hotspot can be widely utilized in fields such as art performance, auxiliary teaching, digital twins. The surface is aimed at exploring how to use projector cast images an surface, then achieve consistent display effect with geometric shape. In pursuit more precise on surfaces, this paper proposes for irregular based...
Background subtraction methods are widely exploited for moving object detection in many applications. How to correctly and efficiently model the background is key technique such approaches. We review typical systems characterized by pixel-wise multiple Gaussians statistics. Our research on initialization introduced afterwards, where a fast online algorithm devised train quickly correctly. Then convenient way combine luminance distortion with chrominance proposed shadow complex scene....
Abstract. In this paper, we propose a method of object detection based on thermal images acquired from unmanned aerial vehicles (UAV). Compared with visible images, have lower requirements for illumination conditions, but they some problems, such as blurred edges and low contrast. To address these to use the saliency map image enhancement attention mechanism detector. YOLOv3 network is trained benchmark BASNet used generate maps images. We fuse their corresponding through pixel-level...
Abstract. This paper presents a method for multiple object tracking (MOT) in video streams. The incorporates the prediction of physical locations people into tracking-by-detection paradigm. We predict trajectories on an estimated ground plane and apply learning-based network to extract appearance features across frames. transforms detected from image space refine trajectories. transform allows objects multi-view images be associated under one coordinate system. Besides, occluded pedestrians...