DeTac: Dual-branch Decoupling of Texture and Force Signals in Vision-Based Tactile Sensors

DeTac is a new dual-branch generative framework that concurrently extracts clean shading and marker images from a single mixed tactile image, addressing challenges in existing data-driven decoupling methods. The approach uses full-image synthesis to bypass mask dependency and introduces cross-branch synergistic optimization for mutual supervision. A real-world paired tactile dataset is constructed to provide high-quality supervision, and experiments show state-of-the-art decoupling accuracy.