A5101959190- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Multimodal Machine Learning Applications
- Image and Object Detection Techniques
- Video Surveillance and Tracking Methods
- Medical Image Segmentation Techniques
- Image Processing Techniques and Applications
- Visual Attention and Saliency Detection
- Remote-Sensing Image Classification
- COVID-19 diagnosis using AI
- Remote Sensing and Land Use
- Digital Holography and Microscopy
- Image Enhancement Techniques
- Mineral Processing and Grinding
- Planetary Science and Exploration
- Image Retrieval and Classification Techniques
- Face and Expression Recognition
- Hand Gesture Recognition Systems
- Human Pose and Action Recognition
- Underwater Acoustics Research
- Face recognition and analysis
- Image Processing and 3D Reconstruction
- Modular Robots and Swarm Intelligence
- Advanced Materials and Mechanics
Technology and Engineering Center for Space Utilization
2019-2024
Chinese Academy of Sciences
2017-2024
Zhejiang University
2018-2024
Tianjin Medical University
2024
Beijing Academy of Artificial Intelligence
2022-2024
University of Chinese Academy of Sciences
2018-2024
Institute of Automation
2020-2024
Tianjin Stomatological Hospital
2024
Fudan University
2021-2024
First Affiliated Hospital of Jinan University
2021-2023
Recently, anchor-free object detectors have shown promising performance in oriented detection on remote sensing images. However, the objects images always large variations arbitrary orientations, sizes, and aspect ratios, which makes existing methods hard to obtain satisfactory results. In this article, we propose a novel detector, center-boundary dual attention (CBDA) network (CBDA-Net), for fast accurate CBDA-Net, construct CBDA module, utilizes mechanism extract features center boundary...
Artificial intelligence (AI) has unparalleled potential to unlock useful information from real-world data innovate trial design. Here, we discuss how AI can be used optimize clinical design and potentially boost the success rate of trials. Zhang et al. artificial leverage valuable insights for innovative
Most object detection methods require huge amounts of annotated data and can detect only the categories that appear in training set. However, reality acquiring massive is both expensive time-consuming. In this paper, we propose a novel two-stage detector for accurate few-shot detection. first stage, employ support-query mutual guidance mechanism to generate more support-relevant proposals. Concretely, on one hand, query-guided support weighting module developed aggregating different supports...
We propose a new model for fast and accurate video object segmentation. It consists of two convolutional neural networks, Dynamic Targeting Network (DTN) Mask Refinement (MRN). DTN locates the by dynamically focusing on regions interest surrounding target object. The region is predicted via sub-streams, Box Propagation (BP) Re-identification (BR). BP stream faster but less effective at objects with large deformation or occlusion. BR performs better in difficult scenarios higher computation...
Object goal visual navigation is a challenging task that aims to guide robot find the target object based on its observation, and limited classes pre-defined in training stage. However, real households, there may exist numerous needs deal with, it hard for all of these be contained To address this challenge, we study zero-shot task, which at guiding robots targets belonging novel without any samples. end, also propose framework called semantic similarity network (SSNet). Our use detection...
Origami structures hold promising potential in space applications, such as ultra-large-area solar arrays, deployable stations, and extra-terrestrial modular foldable buildings. However, the development of thick-panel origami has been limited, relying on a few typical patterns without comprehensive design theory for multi-crease, multi-vertex configurations. Additionally, realizing closed Polyhedra presents substantial challenges. Here, we introduce methodology inspired by kirigami principles...
To achieve accurate and robust object detection in the real-world scenario, various forms of images are incorporated, such as color, thermal, depth. However, multimodal data often suffer from position shift problem, i.e., image pair is not strictly aligned, making one has different positions modalities. For deep learning method, this problem makes it difficult to fuse features puzzles convolutional neural network (CNN) training. In article, we propose a general detector named aligned region...
Green innovation has been attracting increasing attention due to its contributions the conservation of resources and environmental protection. However, in process exploring green innovation, allocation direction are often inaccurate, which leads a low efficiency innovation. If we can learn practices from successful companies, certainly provide reference strategies for those companies that Therefore, taking Fortune Global 500 as analysis object, this research develops criteria conducts...
In recent years, Few-shot Object Detection (FSOD) has become an increasingly important research topic in computer vision. However, existing FSOD methods require strong annotations including category labels and bounding boxes, their performance is heavily dependent on the quality of box annotations. acquiring both expensive time-consuming. This inspires study weakly supervised (WS-FSOD short), which realizes with only image-level annotations, i.e., labels. this paper, we propose a new...
Recently, Faster Region-based CNN(Faster R-CNN) has achieved marvelous accomplishment in object detection and recognition. In this paper, R-CNN is applied to marine organism However, the training of requires a mass labeled samples which are difficult obtain for organism. Therefore, three data augmentation methods proposed dedicated underwater-imaging. Specifically, inverse process underwater image restoration used simulate different turbulence environments. Perspective transformation view...
Current state-of-the-art two-stage detectors heavily rely on region proposals to guide the accurate detection for objects. In previous proposal approaches, interaction between different functional modules is correlated weakly, which limits or decreases performance of approaches. this paper, we propose a novel strong correlation learning framework, abbreviated as SC-RPN, aims set up stronger relationship among in task. Firstly, Light-weight IoU-Mask branch predict intersection-over-union...
The task of zero-shot object goal visual navigation (ZSON) aims to enable robots locate previously "unseen" objects by observations. This presents a significant challenge since the robot must transfer policy learned from "seen" through auxiliary semantic information without training samples, process known as learning. In order address this challenge, we propose novel approach termed Semantic Policy Network (SPNet). SPNet consists two modules that are deeply integrated with embeddings: Actor...