A5076564474- Domain Adaptation and Few-Shot Learning
- Human Pose and Action Recognition
- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Gait Recognition and Analysis
- Multimodal Machine Learning Applications
- Anomaly Detection Techniques and Applications
- Advanced Graph Neural Networks
- Digital Media Forensic Detection
- Infrared Target Detection Methodologies
- Advanced Steganography and Watermarking Techniques
- Remote-Sensing Image Classification
- Image Enhancement Techniques
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Complex Network Analysis Techniques
- Video Analysis and Summarization
- Image Processing Techniques and Applications
- Face and Expression Recognition
- Machine Learning and ELM
- Text and Document Classification Technologies
- Generative Adversarial Networks and Image Synthesis
- Fire Detection and Safety Systems
- Advanced Algorithms and Applications
Xi'an University of Technology
2015-2024
Chang'an University
2005
In this paper, we propose a three-stream convolutional neural network (3SCNN) for action recognition from skeleton sequences, which aims to thoroughly and fully exploit the data by extracting, learning, fusing inferring multiple motion-related features, including 3D joint positions displacements across adjacent frames as well oriented bone segments. The proposed 3SCNN involves three sequential stages. first stage enriches independently extracted features co-occurrence feature learning....
Gesture recognition has been applied in many fields as it is a natural human–computer communication method. However, of dynamic gesture still challenging topic because complex disturbance information and motion information. In this paper, we propose an effective method by fusing the prediction results two-dimensional (2D) representation convolution neural network (CNN) model three-dimensional (3D) dense convolutional (DenseNet) model. Firstly, to obtain compact discriminative representation,...
Due to the explosive increase in online videos, near-duplicate video retrieval (NDVR) has attracted much researcher attention. NDVR very wide applications, such as copyright protection, monitoring, and automatic tagging. Local features serve elementary building blocks many algorithms, most of them exploit local volume information using a bag (BOF) representation. However, representation ignores potentially valuable about global distribution interest points. Moreover, discriminative power...
The classification of marine vessels is one the important problems maritime traffic. To fully exploit complementarity between different features and to more effectively identify vessels, a novel feature structure fusion method based on spectral regression discriminant analysis (SF-SRDA) was proposed. Firstly, we selected convolutional neural network that better describe characteristics ships, constructed graphs by similarity metric. Then weighed concatenate multi-feature fused their...
Effective detection of infrared (IR) moving small targets in complex cluttered environments plays a key role IR search and track systems for self-defense or attacks. In this letter, an small-target algorithm utilizing spatiotemporal consistency motion trajectories is proposed. First, feature points are densely sampled tracked using the dense optical flow to compute trajectories. Second, suspected deleted by characteristics target. Third, under assumption that each target defined as compact...
Recovering high-quality inscription images from unknown and complex noisy is a challenging research issue. Different natural images, character pay more attention to stroke information. However, existing models mainly consider pixel-level information while ignoring structural of the character, such as its edge glyph, resulting in reconstructed with mottled local structure damage. To solve these problems, we propose novel generative adversarial network (GAN) framework based on an edge-guided...
Structure fusion (SF) has been presented for multiple feature via mining the discriminative and complementary information from different sets. As typical methods, SF based on locality preserving projections (SFLPP) tensor subspace analysis (SFTSA) have developed classification by capturing complete structure features. However, jointed optimisation function of SFLPP or SFTSA does not clearly explain modelling mechanism SF, its solving process is complex because iterative eigenvalue...
We study Automatic Target Recognition (ATR) in infrared (IR) imagery from the perspective of feature fusion. The key to fusion is take advantage discriminative and complementary information different sets, which can be represented as internal (within each set) or external structures (across sets). Traditional approaches tend preserve either via certain projection. Some early attempts consider both implicitly indirectly without revealing their relative importance relevance. propose a new...