Class Incremental Semantic

Class incremental semantic segmentation (CISS) focuses on training semantic segmentation models to learn new object classes sequentially without forgetting previously learned ones, a crucial challenge in lifelong machine learning. Current research emphasizes mitigating "catastrophic forgetting" through techniques like knowledge distillation, weight fusion, and generative replay, often employing convolutional neural networks or, increasingly, transformers. These advancements are vital for developing robust and adaptable AI systems in domains like autonomous driving and robotics, where continuous learning from new data is essential.