A5101849245- Machine Learning and Data Classification
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Natural Language Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Real-time simulation and control systems
- Visual Attention and Saliency Detection
- Multimodal Machine Learning Applications
- Machine Learning and Algorithms
University of Macau
2023-2025
In image stitching, artifacts caused by misalignment affect the visual quality and performance of subsequent tasks such as segmentation detection. This paper proposes SMPR, a reconstruction-based aligned composition method to minimize artifacts. SMPR fuses images in part overlapping areas reconstructs other portions from single images. Specifically, we propose seam mask generation obtain optimal masks that pass through minimal misalignment. During training, use guide model detecting fusion...
In semantic segmentation, accurate prediction masks are crucial for downstream tasks such as medical image analysis and editing. Due to the lack of annotated data, few-shot segmentation (FSS) performs poorly in predicting with precise contours. Recently, we have noticed that large foundation model segment anything (SAM) well processing detailed features. Inspired by SAM, propose FSS-SAM boost FSS methods addressing issue inaccurate contour. The is training-free. It works a post-processing...
In computer vision, deep learning models excel in tasks with data adhering to the independently and identically distributed principle but struggle distribution shifts between source target domains. Domain Generalization (DG) aims train on for effective performance unseen This paper addresses two pivotal challenges DG: tendency of overly focus transferability at cost distinguishability, common oversight imbalanced classes long-tailed datasets. We introduce an innovative approach that employs...