A5006091893- Advanced Neural Network Applications
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Gene expression and cancer classification
- Advanced Image and Video Retrieval Techniques
- Vehicle License Plate Recognition
- Industrial Vision Systems and Defect Detection
- Video Surveillance and Tracking Methods
- Remote-Sensing Image Classification
- Handwritten Text Recognition Techniques
- Advanced Image Fusion Techniques
- Image Processing and 3D Reconstruction
- Domain Adaptation and Few-Shot Learning
- Digital Imaging for Blood Diseases
- Face and Expression Recognition
- Robotics and Sensor-Based Localization
- Human Motion and Animation
- Infrared Target Detection Methodologies
- Image Processing Techniques and Applications
- Color Science and Applications
- Optical measurement and interference techniques
China Earthquake Administration
2024
Seismological Bureau of Shanghai
2022
Huaqiao University
2019-2020
The classification of benign and malignant based on ultrasound images is great value because breast cancer an enormous threat to women’s health worldwide. Although both texture morphological features are crucial representations tumor images, their straightforward combination brings little effect for improving the since high-dimensional too aggressive so that drown out low-dimensional features. For that, efficient feature combing method proposed improve malignant. Firstly, (i.e., local binary...
Aerial object detection is a crucial task in computer vision because it plays pivotal role understanding remote images. However, most Convolutional Neural Network (CNN) methods primarily focus on the spatial/channel interactions, overlooking significance of frequency domain information. To overcome these limitations, we introduce an innovative method named Selective Frequency Interaction (SFI) network for aerial detection. Our comprises two essential modules: Frequency-domain Feature...
Vehicle re-identification's methods usually exploit the spatial uniform partition strategy via dividing deep feature maps into several parts. Then each of them is further independently processed by multi-network branch to obtain refined part features. However, cooperation among those features underestimated. This paper proposes a contrastive attention module (CAM) assess one feature's importance based on all Practical derived re-weighting feature. Furthermore, flexible CAM network (CAMNet)...
Fabric image retrieval aims to match a probe fabric from gallery set, playing great role in the textile industry. However, there is lack of one comprehensive benchmark for retrieval. For this, large proposed this paper. Firstly, dataset namely 1.0 proposed, which contains 46,656 images 972 subjects. Each subject collected with multiple various angles and both front back sides. Secondly, evaluation protocol designed evaluate accuracy, uses cumulative characteristic (CMC) curve mean average...
Object re-identification method is made up of backbone network, feature aggregation, and loss function. However, most networks lack a special mechanism to handle rich scale variations mine discriminative representations. In this paper, we firstly design hierarchical similarity graph module (HSGM) reduce the conflict networks. The designed HSGM builds mapping relationships between global-local local-local. Secondly, divide map along with spatial channel directions in each graph. applies...
In this report, we introduce the technical details of our submission to VIPriors object detection challenge. Our solution is based on mmdetction a strong baseline open-source toolbox. Firstly, an effective data augmentation method address lack problem, which contains bbox-jitter, grid-mask, and mix-up. Secondly, present robust region interest (ROI) extraction learn more significant ROI features via embedding global context features. Thirdly, propose multi-model integration strategy...
Breast cancer has become one of the malignant tumors with highest morbidity and mortality among women in world, which seriously threatens women's health. Because cause breast is unknown, early detection, diagnosis treatment are key to improve cure rate cancer. Computer aided (CAD) system a valuable assistant method for automatic detection classification In this paper, morphological texture features ultrasonic images extracted train SVM classifier, so as realize tumors. Of 1061 newly...
The Vehicle Logo Recognition (VLR) is becoming more and important in the Automated Classification (AVC) research area since it can effectively assist vehicle classification. However, existing open-access logo datasets have many issues, which are hard to represent real-world conditions such as occlusions, low image resolutions varying lighting conditions, thus heavily hindering VLR application reality. For this, a large-scale benchmark proposed this paper. Firstly, dataset namely VLD 1.0...
Computer-aided diagnosis (CAD) system for breast ultrasound can effectively assist doctors in classifying benign and malignant of tumors. CAD increases the work efficiency diagnostic accuracy, it also reduces rate misdiagnosis. This paper proposes a method based on histogram oriented gradients (HOG), local binary pattern (LBP) gray-level co-occurrence matrix (GLCM) feature extraction combined with machine learning classifier - support vector (SVM) to classify tumor images. Besides, due lack...