- Geotechnical Engineering and Soil Stabilization
- Geotechnical Engineering and Soil Mechanics
- Climate change and permafrost
- Soil and Unsaturated Flow
- Human Pose and Action Recognition
- Landslides and related hazards
- Human Motion and Animation
- Grouting, Rheology, and Soil Mechanics
- Geotechnical Engineering and Underground Structures
- Natural Language Processing Techniques
- Topic Modeling
- Rough Sets and Fuzzy Logic
- Advanced Image Processing Techniques
- Biomedical Text Mining and Ontologies
- Text and Document Classification Technologies
- Cryospheric studies and observations
- Face recognition and analysis
- Groundwater flow and contamination studies
- 3D Shape Modeling and Analysis
- Traffic Prediction and Management Techniques
- Hand Gesture Recognition Systems
- Geotechnical Engineering and Analysis
- Time Series Analysis and Forecasting
- Advanced Text Analysis Techniques
- Advanced Vision and Imaging
Nanjing Hydraulic Research Institute
2011-2025
Max Planck Institute for Intelligent Systems
2017-2024
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering
2018-2022
Shanghai Jiao Tong University
2021-2022
National Dong Hwa University
2021
The University of Western Australia
2019
Ann Arbor Center for Independent Living
2015
University of Michigan
2013-2015
Harbin Institute of Technology
2015
Changhua Christian Hospital
2015
We demonstrate a novel deep neural network capable of reconstructing human full body pose in real-time from 6 Inertial Measurement Units (IMUs) worn on the user's body. In doing so, we address several difficult challenges. First, problem is severely under-constrained as multiple parameters produce same IMU orientations. Second, capturing data conjunction with ground-truth poses expensive and to do many target application scenarios (e.g., outdoors). Third, modeling temporal dependencies...
Existing markerless motion capture methods often assume known backgrounds, static cameras, and sequence specific priors, limiting their application scenarios. Here we present a fully automatic method that, given multi-view videos, estimates 3D human pose body shape. We take the recently proposed SMPLify \cite{bogo2016keep} as base extend it in several ways. First fit model to 2D features detected images. Second, use CNN segment person each image contours, further improving accuracy. Third...
Existing marker-less motion capture methods often assume known backgrounds, static cameras, and sequence specific priors, which narrows its application scenarios. Here we propose a fully automatic method that given multi-view video, estimates 3D human body shape. We take recent SMPLify \cite{bogo2016keep} as the base method, extend it in several ways. First fit to 2D features detected images. Second, use CNN segment person each image model contours further improves accuracy. Third utilize...
We demonstrate a novel deep neural network capable of reconstructing human full body pose in real-time from 6 Inertial Measurement Units (IMUs) worn on the user's body. In doing so, we address several difficult challenges. First, problem is severely under-constrained as multiple parameters produce same IMU orientations. Second, capturing data conjunction with ground-truth poses expensive and to do many target application scenarios (e.g., outdoors). Third, modeling temporal dependencies...
Freeze–thaw (F-T) cycles are one of the main deterioration factors for performance expansive clay in seasonally frozen soil regions. This study designed a unidirectional environmental boundary loading device F-T cycle tests. Laboratory tests were conducted to quantitatively evaluate effects on volume change behavior and mechanical properties with different degrees compaction below canal Xinjiang. The SEM image processing was performed Image-Pro Plus software get pore parameters. results...
Abstract Humans constantly interact with objects to accomplish tasks. To understand such interactions, computers need reconstruct these in 3D from images of whole bodies manipulating objects, e.g., for grasping, moving and using the latter. This involves key challenges, as occlusion between body motion blur, depth ambiguities, low image resolution hands graspable object parts. make problem tractable, community has followed a divide-and-conquer approach, focusing either only on interacting...
The continuing interest and progress in indigenous communities local economies based on traditional, cultural, ecological knowledge contributes to resilience. Here we report an ongoing collaborative project investigating the process of renewal cultural heritage through strengthening roots traditions practice, changing concepts tradition. investigates various mechanisms for conserving culture: How culture is reconstructed; how this related social frame practice everyday life; power...
Phase change materials (PCMs) can store large amounts of energy in latent heat and release it during phase changes, which could be used to improve the freeze-thaw performance soil. The composite material was prepared with paraffin as PCM 8% Class C fly ash (CFA) supporting material. Laboratory tests were conducted reveal influence (CFA-PCM) on thermal properties, volume changes mechanical properties expansive results show that failed establish a good improvement effect due leakage. CFA...
Cyclic action of wetting–drying (WD) is a main cause for the deterioration and failure expansive soil canals. This paper presents details an investigation dealing with influences WD cycles on characteristics modes canals in Xinjiang, China by laboratory centrifugal model tests. The results show that strength parameters have significant reduction crack development cyclic processes. final attenuation ranges cohesion internal friction angle are about 53%, 46% 37%, 28%, respectively....
Text description of engineering diagnoses recorded during and after vehicle repair process plays an important role in root cause analyzing maintenance. The fact that such text is unstructured, lack grammar, has a lot spelling errors large amount self-invented domain specific terminologies introduces challenges difficulties for automatic information retrieving categorization. This paper presents our research mining diagnostic applications. Specifically, typo correction system proposed...
This study focused on the swelling behavior of rock from canal basement under multiple drying‐wetting (D‐W) cycles. A series laboratory tests were conducted a rock, with cracking and strength behaviors investigated. By using image‐processing technique, crack patterns described, then quantitatively analyzed basis fractal dimension. The experimental data indicated that ability, including level, dimension, strength, decrease increasing drying wetting cycle. On this basis, centrifuge model...
A mathematical model reflecting the evolution of both particle breakage and particle-size distribution (PSD) for rockfill subjected to triaxial shear has been proposed in this article. First, PSD a soil is expressed by gradation equation which shown be widely applicable comparison with large amount test data from soils encountered practice. Then two factors Bg Br are suggested proved suitable model. Mathematical relationships between Bg, equation's parameters b m derived. Then, correlations...
This paper reports on the mechanical behavior of cement solidified dredged material (SDM) tested after 7 and 28 days curing. The stress-strain curves, unconfined compressive strength (UCS), cohesion internal friction angle between SDM crushed (CSDM) were compared analyzed. Test results show that curves change from ductile to brittle as curing time increases. However, CSDM only behavior. UCS increases with increasing time. substantially in a relatively short rate increment reduces period 14...
Large amount of sediments are dredged from rivers and lakes as a result environmental dredging in China. These materials (DM) have poor geotechnical properties normally treated wastes. On the other hand, there is huge demand sub-grade due to increasing number highway construction projects eastern Thus reuse DM material may be considered an environmental-friendly economical option. In this study taken Tai-hu lake was modified by adding fine sand, quicklime Portland cement. Water content,...
Typo detection and correction is an important process in many text mining applications. This research focuses on automatic typo for processing documents that are unstructured, contain grammar spelling errors, have self-invented terminologies can be interpreted only through domain-specific knowledge. In this paper we present intelligent (ITDC) system. Its 'intelligence' reflected by automatically identifying accurately correcting a broad range of typos, from simple typos such as duplication,...