A5025295832- Advanced Vision and Imaging
- Robotics and Sensor-Based Localization
- Image Processing Techniques and Applications
- Advanced Image Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Electrocatalysts for Energy Conversion
- Video Surveillance and Tracking Methods
- Catalytic Processes in Materials Science
- Minerals Flotation and Separation Techniques
- Concrete and Cement Materials Research
- Advanced battery technologies research
- Nanomaterials for catalytic reactions
- Calcium Carbonate Crystallization and Inhibition
- Biometric Identification and Security
- Fuel Cells and Related Materials
- Magnesium Oxide Properties and Applications
- Vehicle License Plate Recognition
- Visual Attention and Saliency Detection
- Layered Double Hydroxides Synthesis and Applications
- Digital Holography and Microscopy
- Image Enhancement Techniques
- Smart Agriculture and AI
- Optical measurement and interference techniques
- Advanced Neural Network Applications
- Indoor and Outdoor Localization Technologies
Shanghai Institute of Microsystem and Information Technology
2018-2021
Hefei University of Technology
2019-2021
University of Chinese Academy of Sciences
2018-2021
State Key Laboratory of Transducer Technology
2018-2021
Chinese Academy of Sciences
2020-2021
Catalytic Materials (United States)
2020
Hefei University
2019-2020
Cu atoms and a hollow structure are helpful for the improvement of MOR activity Pt. Trace Ir might promise superior durability Pt-based catalysts while maintaining catalytic activity.
The 3D semantic map plays an increasingly important role in a wide variety of applications, especially for many kinds task-driven robots. In this paper, we present mapping methodology obtaining from RGB-D scans. contrast to existing methods that use annotated information as supervisory, focus on accurate 2D frame labeling and combine labels space using fusion mechanism. For scene parsing, two-stream network with novel discriminatory mask loss is proposed explore sufficient extraction RGB...
Large quantities of solid waste phosphogypsum (PG) not only occupy the farmland but also cause severe pollution to surrounding environment. The most effective way achieve a high-value utilization PG is prepare α-calcium sulfate hemihydrate (α-HH) with various morphologies. In this study, we developed simple and Cl-free one-pot synthetic strategy effectively transform into single-crystal α-HH regular hexagonal microplates (RHMPs) in Na2SO4 aqueous solution at 98 °C assisted sodium citrate...
Visual Simultaneously Localization and Mapping (SLAM) systems obtain accurate ego-motion of cameras 3D representation the environment. Although performance SLAM is very impressive in many common scenes, tracking failure still a challenging issue, which always exist low-textured environment rapid camera motion situations. In this paper, we propose real-time stereo visual system using bionic eyes. The new type sensor named eyes are basically perform all movements human Our makes good...
This paper presents a semantic-guided interpolation scheme (SemFlow) to handle motion boundaries and occlusions in large displacement optical flow. The basic idea is segment images into superpixels estimate their homographies for interpolation. In order ensure each superpixel can be approximated as plane, refinement method introduced. Moreover, we put forward homography estimation model weighted by the distance between its <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML"...
The induction time of DH crystals is extended by the introduction Cu<sup>2+</sup>.
Scene understanding is one of the foundations for robots to achieve true artificial intelligence. Semantic segmentation that imitates mechanism human visual system can effectively promote correctness scene understanding. It conforms basic principle environment perception and enables better serve society. In this paper, we propose a multilevel cross-aware network (MCA-Net) RGBD semantic segmentation. utilizes residual structure encode texture information depth geometric respectively. view...
Disparity calculation is crucial for binocular sensor ranging. The disparity estimation based on edges an important branch in the research of sparse stereo matching and plays role visual navigation. In this paper, we propose a robust method semantic edges. Some simple costs are used first, then novel adaptive dynamic programming algorithm proposed to obtain optimal solutions. This makes use or consistency constraint between images adaptively search parameters, which can improve robustness...
Visual simultaneously localization and mapping (SLAM) systems obtain accurate estimation of the camera motion 3D map environment. Although performance SLAM is very impressive in many common scenarios, tracking failure still a challenging issue, which always exists low textured environment rapid situation. In this paper, we proposed first active bionic eyes system that leverages saccade movement human eyes. order to find more features points, an autonomous control strategy eyes, mainly...
Microscopic profilometry based on fringe projection is notable for its noncontact and robust nature widely used in industrial electronics. However, ability to observe electronic packages limited by the depth of field (DOF) lens. This article proposes a novel reconstruction scheme telecentric 3-D overcome this limitation. The advantage phase-shifting utilized through defocus detection achieve high-precision contour measurement with single DOF. After completing such local reconstruction,...
Recent CNN-based optical flow approaches have a separated structure of feature extraction and estimation. The core task is finding the corresponding points while rich representation just key part such matching problems. However, prior work usually pays more attention to design decoder than extraction. In this paper, we present novel estimation network enrich each pyramid level, with hierarchical dilated architecture bottom-up aggregation scheme. addition, inspired by edge guided classical...
Learning to distinguish independent moving objects from the observed optical flow with a camera remains challenging. In this work, we first present novel pose compensation (CPC) scheme. With help of ingenious geometric analysis, it breaks into patterns that are easier interpret for motion segmentation network. Secondly, further refine such parameter aware (CPA) module account poses' errors in CPC processing and enhance entire network's tolerance noises. Additionally, an MMPNet is developed...