A5101827786- Video Surveillance and Tracking Methods
- Face recognition and analysis
- Human Pose and Action Recognition
- Gait Recognition and Analysis
- Image Enhancement Techniques
- Advancements in Battery Materials
- Speech and dialogue systems
- Topic Modeling
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Plant Genetic and Mutation Studies
- Advanced Vision and Imaging
- Multimodal Machine Learning Applications
- Advanced Battery Materials and Technologies
- Biometric Identification and Security
- Hand Gesture Recognition Systems
- Image Processing Techniques and Applications
- Plant tissue culture and regeneration
- Advanced Image Processing Techniques
- High voltage insulation and dielectric phenomena
- Computer Graphics and Visualization Techniques
- Plant-Microbe Interactions and Immunity
- Generative Adversarial Networks and Image Synthesis
- Extraction and Separation Processes
Nanjing University of Aeronautics and Astronautics
2020-2024
University of Wollongong
2016
Abstract The objective of human pose estimation (HPE) derived from deep learning aims to accurately estimate and predict the body posture in images or videos via utilization neural networks. However, accuracy real-time HPE tasks is still be improved due factors such as partial occlusion parts limited receptive field model. To alleviate loss caused by these issues, this paper proposes a model called $${\textbf {CCAM-Person}}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML">...
Pristine and boron-doped anatase TiO2 were prepared via a facile sol-gel method the hydrothermal for application as anode materials in sodium-ion batteries (SIBs). The leads to agglomerated TiO2, whereas is conducive formation of highly crystalline discrete nanoparticles. structure, morphology, electrochemical properties studied. crystal size with boron doping smaller than that nondoped crystals, which indicates addition can inhibit growth. measurements demonstrated reversible capacity...
Cross-modality visible-infrared person reidentification (VI-ReID), which aims to retrieve pedestrian images captured by both visible and infrared cameras, is a challenging but essential task for smart surveillance systems. The huge barrier between has led the large cross-modality discrepancy intraclass variations. Most existing VI-ReID methods tend learn discriminative modality-sharable features based on either global or part-based representations, lacking effective optimization objectives....
Super-Resolution (SR) is a fundamental computer vision task that aims to obtain high-resolution clean image from the given low-resolution counterpart. This paper reviews NTIRE 2021 Challenge on Video Super-Resolution. We present evaluation results two competition tracks as well proposed solutions. Track 1 develop conventional video SR methods focusing restoration quality. 2 assumes more challenging environment with lower frame rates, casting spatio-temporal problem. In each competition, 247...
Traditional text-based person re-identification (ReID) techniques heavily rely on fully matched multi-modal data, which is an ideal scenario. However, due to inevitable data missing and corruption during the collection processing of cross-modal incomplete issue usually met in real-world applications. Therefore, we consider a more practical task termed ReID task, where images text descriptions are not completely contain partially modality data. To this end, propose novel Prototype-guided...
Cross-modality visible-Infrared person re-identification (cm-ReID) is extremely challenging due to the huge modality discrepancy between RGB and IR modalities. Existing methods focus on sample features themselves, trying learn modality-invariant perform alignment reduce in dataset-level, while negative impact of specific identity optimization are not specifically addressed. Moreover, most that only extracts appearance cannot acquire enough discriminative matching information for identifying...
In this work we address the problem of rain streak removal with RAW images. The general approach is firstly processing data into RGB images and removing Actually original information in affected by image signal (ISP) pipelines including none-linear algorithms, unexpected noise, artifacts so on. It gains more benefit to directly remove before being processed format. To solve problem, propose a joint solution for obtain clean color from rainy image. be specific, generate converting space...
RGB-Infrared person Re-identification (RGB-IR ReID) aims to match pedestrians with the same identity and different modalities in 24-hour intelligent surveillance system, facing challenge of how mitigate intra-class variations large cross-modality discrepancy. Although achieving promising progress, existing methods still suffer from two main drawbacks. On one hand, conventional struggle aligning RGB infrared into space by simply learning global or partial modality-shared features, which rich...
Text-based person re-identification(ReID) is a text-to-image retrieval task that searches images of people across non-overlapping cameras with textual descriptions. Due to significant intra-class variations and cross-modality disparities, along substantial sample noise, it becomes challenging learn discriminative features. Existing text-based ReID methods tend focus on learning global representations, which possess limited discriminability weak robustness noisy samples. In this paper, we...
Person re-identification (re-ID) aims to establish identity correspondence across different cameras. State-of-the-art re-ID approaches are mainly clustering-based Unsupervised Domain Adaptation (UDA) methods, which attempt transfer the model trained on source domain target domain, by alternatively generating pseudo labels clustering target-domain instances and training network with generated perform feature learning. However, these suffer from problem of inevitable label noise caused...