Yuan Li

ORCID: 0000-0002-7023-8677
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Geomechanics and Mining Engineering
  • Rock Mechanics and Modeling
  • Video Coding and Compression Technologies
  • Biometric Identification and Security
  • Geoscience and Mining Technology
  • Advanced Vision and Imaging
  • Face recognition and analysis
  • Image and Video Quality Assessment
  • Landslides and related hazards
  • Advanced Data Compression Techniques
  • Advanced Image Processing Techniques
  • Advanced Neural Network Applications
  • Face and Expression Recognition
  • Video Surveillance and Tracking Methods
  • Mechanical Behavior of Composites
  • Drilling and Well Engineering
  • Advanced Algorithms and Applications
  • Civil and Geotechnical Engineering Research
  • Domain Adaptation and Few-Shot Learning
  • Tunneling and Rock Mechanics
  • Advanced Image and Video Retrieval Techniques
  • Geotechnical and Geomechanical Engineering
  • Reconstructive Facial Surgery Techniques
  • Geotechnical Engineering and Underground Structures
  • Remote Sensing and Land Use

First Affiliated Hospital of Zhengzhou University
2020-2025

Sun Yat-sen University
2025

Peng Cheng Laboratory
2023-2024

University of Science and Technology Beijing
2015-2024

Peking University
2013-2024

Tianjin University of Science and Technology
2021-2024

Sinopec (China)
2003-2024

Shandong Institute of Commerce & Technology
2023-2024

Xi'an Jiaotong University
2023-2024

National Engineering Research Center for Wheat
2024

10.18653/v1/2024.emnlp-main.342 article EN Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing 2024-01-01

Neural architecture search (NAS) has dramatically advanced the development of neural network design. We revisit space design in most previous NAS methods and find number widths blocks are set manually. However, block counts determine scale (depth width) make a great influence on both accuracy model cost (FLOPs/latency). In this paper, we propose to by designing densely connected space, i.e., DenseNAS. The new is represented as dense super network, which built upon our designed routing...

10.1109/cvpr42600.2020.01064 article EN 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2020-06-01

10.1109/cvpr52733.2024.01300 article EN 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024-06-16

10.1016/s1006-706x(11)60096-4 article EN Journal of Iron and Steel Research International 2011-08-01

The capabilities of biometric systems have recently made extraordinary leaps by the emergence deep learning. However, due to lack enough training data, applications neural network in ear recognition filed run into bottleneck. Moreover, effect fine-tuning from some pre-trained models is far less than expected diversity among different tasks. Therefore, authors propose a large-scale database and explore robust convolutional (CNN) architecture for feature representation. images this...

10.1049/iet-bmt.2017.0176 article EN IET Biometrics 2018-01-23

Multi-view pedestrian detection aims to predict a bird's eye view (BEV) occupancy map from multiple camera views. This task is confronted with two challenges: how establish the 3D correspondences views BEV and assemble information across In this paper, we propose novel Stacked HOmography Transformations (SHOT) approach, which motivated by approximating projections in world coordinates via stack of homographies. We first construct transformations for projecting ground plane at different...

10.1109/iccv48922.2021.00599 article EN 2021 IEEE/CVF International Conference on Computer Vision (ICCV) 2021-10-01

Abstract Segmentation of a complete set teeth from three-dimensional (3D) intra-oral scanner images is crucial step in tooth identification procedures. In large-scale disasters with many victims, are often the preferred and reliable source for victim due to their hard non-deformable characteristics. this paper we present study on automatic segmentation images. We propose method based an improved PointNet++ architecture. To address problem inadequate capability teeth-gingival boundary...

10.1007/s44267-023-00026-7 article EN cc-by Visual Intelligence 2023-10-10

Abstract Background Intestinal ultrasound (IUS) is becoming a standard evaluation tool for ulcerative colitis (UC). Since the limited diagnostic efficacy of individual parameters, we aimed to develop and validate predictive model endoscopic disease activity in based on transabdominal intestinal combination with biochemical indices, improve non-invasive UC synthetically Methods This study enrolled 101 consecutive adult patients First Affiliated Hospital Zhengzhou University from June 2022...

10.1093/ecco-jcc/jjae190.0647 article EN Journal of Crohn s and Colitis 2025-01-01

Current research on soil–structure interface properties mainly focuses sand, clay, and silt, with little attention given to sandy gravel. In order study the effects of relative density materials shear behavior gravel–structure interface, a series large-scale direct tests gravel were carried out, stress–strain relationships, volume change curves, strengths investigated. The results show that angle internal friction increases linearly (R2 is 0.998), from 43.0° 48.0° when 0.3 0.9. growth trend...

10.3390/buildings15040546 article EN cc-by Buildings 2025-02-11

Abstract In this article, the contemporary stress state of Zhao–Ping metallogenic belt in eastern China was revealed using overcoring and hydraulic fracturing data, relation between field geological tectonics discussed, stability regional faults under present-day environment evaluated. The results indicate that level is considerably high, distribution intensity uneven. regime primarily characterized by σ H > v h . orientation well-oriented WNW–ESE, which roughly identical to other...

10.1007/s40789-025-00769-2 article EN cc-by International Journal of Coal Science & Technology 2025-03-15

Recently, the Turing test has been used to investigate whether machines have intelligence similar humans. Our study aimed assess ability of an artificial (AI) system for spine tumor detection using test.Our retrospective data included 12179 images from 321 patients developing AI systems and 6635 187 test. We utilized a deep learning-based with Faster R-CNN architecture, which generates region proposals by Region Proposal Network in first stage corrects position size bounding box lesion area...

10.3389/fonc.2022.814667 article EN cc-by Frontiers in Oncology 2022-03-11
Coming Soon ...