A5100766399- Domain Adaptation and Few-Shot Learning
- Smart Agriculture and AI
- Advanced Image and Video Retrieval Techniques
- Remote-Sensing Image Classification
- Advanced Neural Network Applications
- Remote Sensing and Land Use
- Advanced Image Fusion Techniques
- Spectroscopy and Chemometric Analyses
- Multimodal Machine Learning Applications
- Remote Sensing and LiDAR Applications
- Robotic Path Planning Algorithms
- Distributed and Parallel Computing Systems
- Image Enhancement Techniques
- Cloud Computing and Resource Management
- Leaf Properties and Growth Measurement
- Video Analysis and Summarization
- Remote Sensing in Agriculture
- Human Pose and Action Recognition
- Distributed systems and fault tolerance
- Robotics and Sensor-Based Localization
- Parallel Computing and Optimization Techniques
- Distributed Control Multi-Agent Systems
- Horticultural and Viticultural Research
- Image Processing Techniques and Applications
- AI-based Problem Solving and Planning
Beihang University
2024
Alibaba Group (China)
2023
Southeast University
2022
Northwestern Polytechnical University
2021
Guangdong University of Technology
2021
Guangdong Institute of Intelligent Manufacturing
2021
Chongqing University of Technology
2020
Northeastern University
2020
Tsinghua University
2018
Data in the real world tends to exhibit a long-tailed label distribution, which poses great challenges for training of neural networks visual recognition. Existing methods tackle this problem mainly from perspective data quantity, i.e., number samples each class. To be specific, they pay more attention tail classes, like applying larger adjustments logit. However, process, quantity and difficulty are two intertwined equally crucial problems. For some features their instances distinct...
Few-shot classification of remote sensing images has attracted attention due to its important applications in various fields. The major challenge few-shot image scene is that limited labeled samples can be utilized for training. This may lead the deviation prototype feature expression, and thus performance will impacted. To solve these issues, a calibration with feature-generating model proposed classification. In framework, encoder self-attention developed reduce influence irrelevant...
Few-shot remote sensing image scene classification has gained more attention due to its ability recognize novel categories with several annotated samples. However, it is a great challenge extract category characteristics insufficient labeled To address this issue, an iterative distribution learning network (IDLN) proposed for the of few-shot images. Specifically, model cyclic architecture, which composed three modules enhance performance. In each iteration, similarity module calculate...
We propose a deep learning method for multi-focus image fusion.Unlike most existing pixel-level fusion methods, either in spatial domain or transform domain, our directly learns an end-to-end fully convolutional two-stream network.The framework maps pair of different focus images to clean version, with chain layers, layer and deconvolutional layers.Our model has advantages efficiency robustness, yet demonstrates state-of-art quality.We explore parameter settings achieve trade-offs between...
Few-shot classification of hyperspectral image (HSI) has been increasingly abstracted attention due to its superiority adopting new HSI with only a few labeled data available. However, insufficient feature expression still bothers the improvement performance. To address this issue, deep self-attention and mutual-attention few-shot learning (SMA-FSL) method is proposed for classification. Specifically, 3D convolutional embedding network utilized extract spectral-spatial at first. Then, are...
Robust semantic labeling of high-resolution remote sensing images in foggy conditions is crucial for automatic monitoring land covers. This remains a challenging task owing to the low inter-class differentiation yet high intra-class variance and geometric size diversity. Although conventional Convolutional Neural Networks have demonstrated state art performance segmentation, most networks are primarily concerned with standard accuracy, while influence on robustness rarely explored. letter...
In this paper, a mathematical model is established based on the characteristics of task assignment swarm drones. The simulated annealing algorithm introduced into genetic multiple groups. improved applied to solve in drones task. Experiments result show that allocation calculation time base significantly reduced compared with standard algorithm.
The rapid development of smart orchards is conducive to scientific planting and management, the estimation fruit maturity key harvest in orchards. Nowadays, research on bayberry almost nothing, order quickly accurately estimate orchards, a algorithm proposed based multi-feature fusion by machine vision. Firstly, considering local global texture characteristics appearance, image features were extracted GLCM LBP. Simultaneously R, G, B, H S components RGB HSV color space, transformed histogram...
Accelerators such as Field Programmable Gate Arrays (FPGAs) are increasingly used in high performance computing, and the problems they applied to process larger amounts of data. FPGA manufacturers have added new types memory on chip help ease bottleneck; however, burden is designer determine how data allocated different types. We study use ultraRAM for a graph application running Amazon Web Services (AWS) that generates large amount intermediate not subsequently accessed sequentially....