- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Image and Signal Denoising Methods
- Context-Aware Activity Recognition Systems
- Network Security and Intrusion Detection
- Autonomous Vehicle Technology and Safety
- Multimodal Machine Learning Applications
- Gait Recognition and Analysis
- Advanced Image Processing Techniques
- Video Analysis and Summarization
- Image Processing Techniques and Applications
- Caching and Content Delivery
- Sleep and Work-Related Fatigue
- Manufacturing Process and Optimization
- Mobile Ad Hoc Networks
- Network Traffic and Congestion Control
- EEG and Brain-Computer Interfaces
- Generative Adversarial Networks and Image Synthesis
- Complex Network Analysis Techniques
- Soil Geostatistics and Mapping
- Cloud Data Security Solutions
- Advanced Clustering Algorithms Research
- Metaheuristic Optimization Algorithms Research
Northeastern University
2025
Nanyang Technological University
2024
Shandong University
2020-2024
Wuhan Donghu University
2024
Nanjing University of Science and Technology
2023
National University of Singapore
2023
Zhengzhou University of Light Industry
2021-2022
Henan University of Technology
2022
Northeast Normal University
2021
PLA Information Engineering University
2021
Human behavior understanding with unmanned aerial vehicles (UAVs) is of great significance for a wide range applications, which simultaneously brings an urgent demand large, challenging, and comprehensive benchmarks the development evaluation UAV-based models. However, existing have limitations in terms amount captured data, types data modalities, categories provided tasks, diversities subjects environments. Here we propose new benchmark - UAV-Human human UAVs, contains 67,428 multi-modal...
Video Anomaly detection, aiming to detect the abnormal behaviors in surveillance videos, is a challenging task since anomalous events are diversified and complicated different situations. And this makes it difficult use one single static network architecture extract useful information from diverse patterns. Therefore, article, we propose novel Dynamic Self-Supervised Network (DSS-Net) explore both spatial temporal information. In our DSS-Net, design dynamic adaptively select suitable latent...
Early activity prediction/recognition aims to recognize action categories before they are fully conveyed. Compared full-length sequences, partial video sequences only provide insufficient discrimination information, which makes predicting the class labels for some similar activities challenging, especially when very few frames can be observed. To address this challenge, in paper, we propose a novel meta negative network, namely, Magi-Net, that utilizes contrastive learning scheme alleviate...
Fine-grained image classification is a challenging problem because of its large intra-class differences and low inter-class variance. Bilinear pooling based models have been shown to be effective at fine-grained classification, while most previous approaches neglect the fact that distinctive features or modeling distinguishing regions usually an important role in solving problem. In this paper, we propose novel convolutional neural network framework, i.e., attention bilinear pooling, for...
Early activity prediction, which aims to recognize class labels before actions are fully performed, is a very challenging task since partially observed action sequences contain insufficient class-discrimination information, and thus, many partial belonging different categories may look similar. Therefore, in this paper, we propose novel <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">guidance aware network (GA-Net)</i> boost the ability...
Action recognition in video understanding is a challenging task, largely because of the complexity and difficulty temporal modeling, making it suffer from motion information loss misalignment attention spatial dimensions. To overcome these difficulties, we propose novel modeling method called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Adjoint Enhancement Network</i> (AE-Net), which can fully explore clues time long-range structure. The...
Human behavior understanding with unmanned aerial vehicles (UAVs) is of great significance for a wide range applications, which simultaneously brings an urgent demand large, challenging, and comprehensive benchmarks the development evaluation UAV-based models. However, existing have limitations in terms amount captured data, types data modalities, categories provided tasks, diversities subjects environments. Here we propose new benchmark - UAVHuman human UAVs, contains 67,428 multi-modal...
BACKGROUND: High-resolution (HR) magnetic resonance imaging (MRI) provides rich pathological information which is of great significance in diagnosis and treatment brain lesions. However, obtaining HR MRI images comes at the cost extending scan time using sophisticated expensive instruments. OBJECTIVE: This study aims to reconstruct from low-resolution (LR) by developing a deep learning based super-resolution (SR) method. METHODS: We propose feedback network with self-attention mechanism...
Firstly, the application status and problems of machine learning in modern intelligent agricultural production were introduced. A smart monitoring system design based on Internet Things technology algorithm was proposed. sensor node for collecting environment information is designed, a wireless network composed ZigBee nodes used to sense detect information. The software developed by Qt can manage display Secondly, decision constructed using tree analyze problem. communication method...
Early action prediction aiming to recognize which classes the actions belong before they are fully conveyed is a very challenging task, owing insufficient discrimination information caused by domain gaps among different temporally observed domains. Most of existing approaches focus on using temporal domains "guide" partially while ignoring discrepancies between harder low-observed and easier highly The recognition models tend learn samples from may lead significant performance drops...
Driver action recognition has significantly advanced in enhancing driver-vehicle interactions and ensuring driving safety by integrating multiple modalities, such as infrared depth. Nevertheless, compared to RGB modality only, it is always laborious costly collect extensive data for all types of non-RGB modalities car cabin environments. Therefore, previous works have suggested independently learning each fine-tuning a model pre-trained on videos, but these methods are less effective...
Associating driver attention with driving scene across two fields of views (FOVs) is a hard cross-domain perception problem, which requires comprehensive consideration cross-view mapping, dynamic analysis, and status tracking. Previous methods typically focus on single view or map to the via estimated gaze, failing exploit implicit connection between them. Moreover, simple fusion modules are insufficient for modeling complex relationships views, making information integration challenging. To...
Recently, large-scale pre-trained vision-language models (e.g., CLIP), have garnered significant attention thanks to their powerful representative capabilities. This inspires researchers in transferring the knowledge from these large other task-specific models, e.g., Video Action Recognition (VAR) via particularly leveraging side networks enhance efficiency of parameter-efficient fine-tuning (PEFT). However, current approaches VAR tend directly transfer frozen action recognition with minimal...
The package size for Application-Specific Integrated Circuit (ASIC) becomes larger; the mainstream is above 50mm * current communication and networking ASICs with more than 2000 IOs. It will take time to assign logical connections from chip solder bump ball across substrate before layout design start, so an effective method of automatic assignment key turn-around (TAT) reduction. benefit TAT reduction drive fast market based upon chip-package co-design methodology. An example real ASIC in...
Spatial autoregressive model, introduced by Clif and Ord in 1970s has been widely applied many areas of science econometrics such as regional economics, public finance, political sciences, agricultural environmental studies transportation analyses. As information technology grows rapidly, observations are seldom independent from others so a space models can take this dependence into account draw more reliable conclusions between covariates the target variable itself. Based on classical...
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL Source Recursive Expected Utility on Rich Mixture Sets 15 Pages Posted: 13 Jul 2022 See all articles by Soo Hong ChewSoo ChewSouthwestern University of Finance and Economics; National SingaporeGavin KaderSouthwestern Economics (SWUFE)Wenqian WangHong Kong Science Technology (Guangzhou) Date Written: July 4, Abstract Building the Herstein-Milnor...
In order to eliminate the asymmetric projection error, this paper proposes a method calculate point of circle center by use concentric circles. Firstly, basic principles and mathematical model are introduced briefly. Next, take arbitrary secants on circle, relationship between secant midpoint infinity imaging plane is obtained based properties projective geometry. The equations which contain parameters midpoints established according geometric constraints. Then coordinate derived from...