A5100599637- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Robotics and Sensor-Based Localization
- Energy Efficiency and Management
- IoT-based Smart Home Systems
- Multimodal Machine Learning Applications
- Human Motion and Animation
- Curcumin's Biomedical Applications
- Inhalation and Respiratory Drug Delivery
- Digital Media Forensic Detection
- Geological and Geochemical Analysis
- Down syndrome and intellectual disability research
- Energy, Environment, Economic Growth
- Generative Adversarial Networks and Image Synthesis
- Fire Detection and Safety Systems
- Remote Sensing in Agriculture
- Impact of AI and Big Data on Business and Society
- 3D Shape Modeling and Analysis
- Geochemistry and Geologic Mapping
- Transportation Planning and Optimization
- Urban Heat Island Mitigation
- Social Robot Interaction and HRI
- Environmental Sustainability in Business
Liaoning University
2025
Hong Kong University of Science and Technology
2024
University of Hong Kong
2024
Zhejiang University
2020-2024
University of California, Santa Cruz
2023-2024
Craft Engineering Associates (United States)
2023
National Institute Of Veterinary Epidemiology And Disease Informatics
2019-2020
University at Albany, State University of New York
2019
Harbin Engineering University
2018
Visual tracking problem demands to efficiently perform robust classification and accurate target state estimation over a given at the same time. Former methods have proposed various ways of estimation, yet few them took particularity visual itself into consideration. Based on careful analysis, we propose set practical guidelines for high-performance generic object tracker design. Following these guidelines, design our Fully Convolutional Siamese tracker++ (SiamFC++) by introducing both...
The Vision Meets Drone (VisDrone2019) Single Object Tracking challenge is the second annual research activity focusing on evaluating single-object tracking algorithms drones, held in conjunction with International Conference Computer (ICCV 2019). VisDrone-SOT2019 Challenge goes beyond its VisDrone-SOT2018 predecessor by introducing 25 more challenging sequences for long-term tracking. We evaluate and discuss results of 22 participating 19 state-of-the-art trackers collected dataset. are...
Down syndrome is one of the most common genetic disorders. The distinctive facial features provide an opportunity for automatic identification. Recent studies showed that recognition technologies have capability to identify However, there a paucity on identification with technologies, especially using deep convolutional neural networks. Here, we developed method utilizing images and networks, which quantified binary classification problem distinguishing subjects from healthy based...
Urban areas account for more than 70% of fossil fuel carbon dioxide (CO2) emissions worldwide. Recent (OCO-3 released in 2019) and forthcoming (CO2M, TANSAT-2, and GOSAT-GW) greenhouse gas satellites can observe wide area column average dry air mole fraction (XCO2) entire urban areas. Although top-down emission monitoring has improved terms spatial coverage frequency, the challenge remains how to utilize space-based observations perform accurate...
3D perception based on the representations learned from multi-camera bird's-eye-view (BEV) is trending as cameras are cost-effective for mass production in autonomous driving industry. However, there exists a distinct performance gap between BEV and LiDAR object detection. One key reason that captures accurate depth other geometry measurements, while it notoriously challenging to infer such information merely image input. In this work, we propose boost representation learning of student...
Abstract The external attraction of urban functions refers to the ability specific functional areas attract foreign populations, which can reflect importance in regional or even global scope. Existing research tends discuss at macro level, is insufficient for understanding internal city functions. Analyzing from perspective travel characteristics reveal distribution and preferences population, help rational planning In this study, we combine Points interest (POI) signaling data identify...
Visual tracking problem demands to efficiently perform robust classification and accurate target state estimation over a given at the same time. Former methods have proposed various ways of estimation, yet few them took particularity visual itself into consideration. After careful analysis, we propose set practical guidelines for high-performance generic object tracker design. Following these guidelines, design our Fully Convolutional Siamese tracker++ (SiamFC++) by introducing both...
We present SplattingAvatar, a hybrid 3D representation of photorealistic human avatars with Gaussian Splatting embedded on triangle mesh, which renders over 300 FPS modern GPU and 30 mobile device. disentangle the motion appearance virtual explicit mesh geometry implicit modeling Splatting. The Gaussians are defined by barycentric coordinates displacement as Phong surfaces. extend lifted optimization to simultaneously optimize parameters while walking mesh. SplattingAvatar is humans where...
Although neural rendering has made significant advancements in creating lifelike, animatable full-body and head avatars, incorporating detailed expressions into avatars remains largely unexplored. We present DEGAS, the first 3D Gaussian Splatting (3DGS)-based modeling method for with rich facial expressions. Trained on multiview videos of a given subject, our learns conditional variational autoencoder that takes both body motion expression as driving signals to generate maps UV layout. To...
Nighttime light (NTL) data is recognized as a reliable proxy for measuring the scope and intensity of human activity, finding wide application in studies such urbanization monitoring, socioeconomic estimation, ecological environment assessment. However, substantial discrepancies limited temporal coverage existing NTL datasets have constrained their potential long-term research applications. To address this, Light U-Net super-resolution network proposed cross-sensor calibration between...
3D perception based on the representations learned from multi-camera bird's-eye-view (BEV) is trending as cameras are cost-effective for mass production in autonomous driving industry. However, there exists a distinct performance gap between BEV and LiDAR object detection. One key reason that captures accurate depth other geometry measurements, while it notoriously challenging to infer such information merely image input. In this work, we propose boost representation learning of student...
Explicit structural inference is one key point to improve the accuracy of scene parsing. Meanwhile, adversarial training method able reinforce spatial contiguity in output segmentations. To take both advantages learning and simultaneously, we propose a novel deep network architecture called Structural Inference Embedded Adversarial Networks (SIEANs) for pixel-wise labeling. The generator our SIEANs, designed parsing network, makes full use convolutional neural networks long short-term memory...