4 papers - avg viability 6.8
A mask-free AI framework for adaptive video subtitle removal that achieves end-to-end inference and strong zero-shot generalization across multiple languages.
Kiwi-Edit offers precise instruction-based video editing with reference guidance, supported by a new scalable dataset and codebase.
Revolutionizing video matting with zero-shot generalization and a new large-scale dataset, enabling high-quality matte creation for diverse real-world videos.
SAMA enables precise, instruction-based video editing using advanced semantic and motion alignment techniques.