A5011080373- Anomaly Detection Techniques and Applications
- Domain Adaptation and Few-Shot Learning
- Emotion and Mood Recognition
- Industrial Vision Systems and Defect Detection
- COVID-19 diagnosis using AI
- AI in cancer detection
- Occupational Health and Safety Research
- Advanced Malware Detection Techniques
- Image Processing and 3D Reconstruction
- Network Security and Intrusion Detection
- Modular Robots and Swarm Intelligence
- Robotics and Sensor-Based Localization
- Computational Physics and Python Applications
- Face recognition and analysis
- Face and Expression Recognition
- Advanced Neural Network Applications
Xiamen University
2023-2024
Jiangxi University of Science and Technology
2019
Anomaly detection (AD) in 3D point clouds is crucial a wide range of industrial applications, especially various forms precision manufacturing. Considering the demand for reliable AD, several methods have been developed. However, most these approaches typically require training separate models each category, which memory-intensive and lacks flexibility. In this paper, we propose novel Point-Language model with dual-prompts ANomaly dEtection (PLANE). The approach leverages multi-modal prompts...
Anomaly detection plays an essential role in large-scale industrial manufacturing. However, reconstruction-based anomaly methods, as one of the mainstream are prone to incorrectly detecting background noise anomalous regions. Therefore, inspired by multi-task learning, we propose Implicit Foreground-guided Network (IFgNet), which consists a Multi-Task Attention Shared (MTAS) sub-network and discriminative sub-network. Specifically, MTAS implements foreground reconstruction tasks within...
Focusing on the facial-based depression recognition where feature distribution could be shifted due to unlimited variations in facial image acquisition, we propose a novel Low-rank constrained latent Domain Adaptation Depression Recognition (LDADR) framework by jointly utilizing appearance and dynamics features. Under this framework, alleviate domain bias recognition, devote uncover compact more informative space representation minimize divergence as well share discriminative structures...
Anomaly Detection (AD) on medical images enables a model to recognize any type of anomaly pattern without lesion-specific supervised learning. Data augmentation based methods construct pseudo-healthy by "pasting" fake lesions real healthy ones, and network is trained predict in manner. The lesion can be found difference between the unhealthy input output. However, using only manually designed fail approximate irregular lesions, hence limiting generalization. We assume exploring intrinsic...
Focusing on cross-dataset Automated Depression Recognition (ADR) by jointly exploring facial appearance and dynamics feature representations, we explore to propose a novel Latent Domain Adaptation (latDADR) framework via discovering discriminative domain-invariant subspace. In this subspace, the between-domain distribution discrepancy would be semantically minimized, meanwhile within-domain geometric structures also discriminatively preserved. latDADR, respectively optimize two target...