Abstract: We propose a controllable visual enhancer, named DDBF, which is based on cross-modal conditional adversarial learning and aims to dispel darkness and achieve better visible and infrared ...