COURSE OVERVIEW

 This practical and highly-interactive course includes real-life case studies and exercises where participants will be engaged in a series of interactive small groups and class workshops.  

This course is designed to provide participants with a detailed and up-to-date overview of Semantic Segmentation. It covers the improtance of semantic segmentation and the difference between image classification, object detection and segmentation; the types of image segmentation, fundamentals of image processing for segmentation and classical computer vision-based segmentation; the deep learning-based segmentation and apply evaluation metrics for semantic segmentation; the fully convolutional networks (FCN), U-Net architecture, contracting and expanding paths in U-Net; and the skip connections for preserving spatial information and applications of U-Net beyond medical imaging. 

 

Further, the course will also discuss the deeplab models, atrous (dilated) convolutions for enlarged receptive fields and fully connected conditional random fields (CRFs) for refinement; the performance comparison with U-Net and FCN; the pyramid scene parsing network (PSPNet), combining detection and segmentation, attention mechanisms in segmentation and transformer-based semantic segmentation models; the multi-scale and context-aware segmentation and real-time semantic segmentation; the imbalanced data in semantic segmentation; the semantic segmentation in medical imaging, autonomous vehicles, autonomous vehicles and aerial and satellite imaging; and the model compression and deployment and GANs and semi-supervised learning for segmentation. 

During this interactive course, participants will learn the imbalanced data in semantic segmentation; the semantic segmentation in medical imaging, autonomous vehicles and aerial and satellite imaging; the model compression and deployment, GANs and semi-supervised learning for segmentation; the AI and federated learning for privacy-preserving segmentation and advances in 3D semantic segmentation for AR/VR; thesynthetic data for training robust segmentation models and handling privacy concerns in real-world applications; the explainability and transparency in AI predictions; and the best practices for responsible AI development.