Researchers have introduced RestoreMore, a novel framework designed to enable image restoration models to continually expand their capabilities without losing previously learned skills. This approach freezes the original pretrained model as a stable anchor and trains new, residual modules for emerging restoration needs. RestoreMore utilizes a bi-level routing mechanism that identifies relevant restoration capabilities and selects complementary degradation experts, allowing new tasks to leverage existing knowledge and enriching the model's expert bank over time. Experiments across various benchmarks show that RestoreMore effectively acquires new abilities while maintaining and even enhancing its existing ones. AI
IMPACT Enables more adaptable and efficient AI model development by allowing continuous learning without performance degradation.
RANK_REASON The cluster contains a research paper detailing a new framework for AI models. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
