A5100783159- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Multimodal Machine Learning Applications
- Anomaly Detection Techniques and Applications
- Advanced Image and Video Retrieval Techniques
- Metaheuristic Optimization Algorithms Research
- Evolutionary Algorithms and Applications
- Advanced Sensor and Energy Harvesting Materials
- Autonomous Vehicle Technology and Safety
- Advanced Multi-Objective Optimization Algorithms
- COVID-19 diagnosis using AI
- Face recognition and analysis
- Innovative Energy Harvesting Technologies
- Visual Attention and Saliency Detection
- Multilevel Inverters and Converters
- Acoustic Wave Resonator Technologies
- Context-Aware Activity Recognition Systems
- Ferroelectric and Piezoelectric Materials
- Energy Harvesting in Wireless Networks
- Ecology and Conservation Studies
- Industrial Vision Systems and Defect Detection
- Face Recognition and Perception
- Microgrid Control and Optimization
Kyungpook National University
2005-2024
Korea University
2013-2024
Korea Institute of Science and Technology
2021-2023
Sungkyunkwan University
2019-2021
Hanbuk University
2006
We introduce a new neural network-based continual learning algorithm, dubbed as Uncertainty-regularized Continual Learning (UCL), which builds on traditional Bayesian online framework with variational inference. focus two significant drawbacks of the recently proposed regularization-based methods: a) considerable additional memory cost for determining per-weight regularization strengths and b) absence gracefully forgetting scheme, can prevent performance degradation in tasks. In this paper,...
Optimization of the formulation and processability 3D-printable ceramic polymer composites offers a solution to piezoelectric materials with high printability responses. Our approach is based on both composite auxetic structural design. The optimal exhibited strong interfacial adhesion, dispersion stability, low viscosity, smooth surface, resulting in piezoelectricity. In addition, processing parameters, such as intensity application time UV laser, were optimized for processability....
In this paper, we propose a novel method for unusual human activity detection in crowded scenes. Specifically, rather than detecting or segmenting humans, devised an efficient method, called motion influence map, representing activities. The key feature of the proposed map is that it effectively reflects characteristics movement speed, direction, and size objects subjects their interactions within frame sequence. Using further developed general framework which can detect both global local...
Acoustic energy transfer using ferroelectrically augmented triboelectric receivers can efficiently deliver to implantable medical devices, marine cable operation sensors, and electronic devices with electromagnetic interference shielding cases.
Fairness is becoming an increasingly crucial issue for computer vision, especially in the human-related decision systems. However, achieving algorithmic fairness, which makes a model produce indiscriminative outcomes against protected groups, still unresolved problem. In this paper, we devise systematic approach reduces biases via feature distillation visual recognition tasks, dubbed as MMD-based Fair Distillation (MFD). While technique has been widely used general to improve prediction...
Direct air capture (DAC) shows considerable promise for the effective removal of CO
Hybrid energy harvesters, using multiple harvesting mechanisms, have been proposed to overcome the limitations of single-mode harvesters. However, since most hybrid harvesters merely combined generated by each mechanism, they could not synergistically enhance output beyond a simple additive approach. In particular, although thermoelectric-piezoelectric explored for simultaneously utilizing ambient thermal–mechanical flows, conventional design achieve expected sum separately harnessed...
For high-power transducer applications, piezoelectric materials must satisfy a high mechanical quality factor (Qm), strain coefficient (dij), and Curie phase transition temperature (Tc). In this study, we report the structural, piezoelectric, characteristics of Mn-doped 0.15 Pb(Yb1/2Nb1/2)O3–0.48 Pb(Mg1/3Nb2/3)O3–0.37PbTiO3 (0.15PYN–0.48PMN–0.37 PT) ceramics. Mn-doping increased tetragonal fraction 0.15PYN–0.48PMN–0.37 PT ceramics formed oxygen vacancies, resulting in tremendous increase Qm...
Autonomous driving is a safety-critical application that requires high-level understanding of computer vision with real-time inference. In this study, we focus on the computational efficiency an important factor by improving running time and performing multiple tasks simultaneously for practical applications. We propose fast accurate multi-task learning-based architecture joint segmentation drivable area, lane line, classification scene. An encoder–decoder efficiently handles input frames...
With the advances in Unmanned Aerial Vehicles (UAVs) technology, aerial images with huge variations appearance of objects and complex backgrounds have opened a new direction work for researchers. The task semantic segmentation becomes more challenging when capturing inherent features global local context UAV images. In this paper, we proposed transformer-based encoder-decoder architecture to address issue precise feature representation is exploited encoder network using self-attention-based...
Modeling spatiotemporal brain dynamics from high-dimensional data, such as functional Magnetic Resonance Imaging (fMRI), is a formidable task in neuroscience. Existing approaches for fMRI analysis utilize hand-crafted features, but the process of feature extraction risks losing essential information scans. To address this challenge, we present SwiFT (Swin 4D Transformer), Swin Transformer architecture that can learn directly volumes memory and computation-efficient manner. achieves by...
Modern data augmentation strategies such as Cutout, Mixup, and CutMix, have achieved good performance in image recognition tasks. Particularly, the approaches, Mixup that mix two images to generate a mixed training image, could generalize convolutional neural networks better than single image-based approaches Cutout. We focus on fact can improve generalization ability, we wondered if it would be effective apply image. Consequently, propose new method produce self-mixed based saliency map,...
Visual perception is a critical task for autonomous driving. Understanding the driving environment in real time can assist vehicle safely. In this study, we proposed multi-task learning framework simultaneous traffic object detection, drivable area segmentation, and lane line segmentation an efficient way. Our network encoder extracts features from input image three decoders at multilevel branches handle specific tasks. The share feature maps with more similar tasks joint semantic...
Remotely captured images possess an immense scale and object appearance variability due to the complex scene. It becomes challenging capture underlying attributes in global local context for their segmentation. Existing networks struggle inherent features cluttered background. To address these issues, we propose a remote sensing image segmentation network, RSSGLT, semantic of images. We by leveraging benefits transformer convolution mechanisms. RSSGLT is encoder–decoder design that uses...