Search papers, labs, and topics across Lattice.
The paper addresses limitations in single image reflection removal (SIRR) stemming from semantic gaps between pre-trained features and reflection removal features, as well as inconsistencies in reflection labels between synthetic and real data. To mitigate these issues, the authors introduce GFRRN, which incorporates parameter-efficient fine-tuning (PEFT) with Mona layers for feature alignment and a label generator to unify reflection labels. GFRRN also employs a Gaussian-based Adaptive Frequency Learning Block (G-AFLB) and a Dynamic Agent Attention (DAA) mechanism, achieving state-of-the-art performance in SIRR.
Achieve superior single image reflection removal by closing the semantic gap between pre-trained models and reflection removal models, and unifying reflection labels across synthetic and real-world data.
Prior dual-stream methods with the feature interaction mechanism have achieved remarkable performance in single image reflection removal (SIRR). However, they often struggle with (1) semantic understanding gap between the features of pre-trained models and those of reflection removal models, and (2) reflection label inconsistencies between synthetic and real-world training data. In this work, we first adopt the parameter efficient fine-tuning (PEFT) strategy by integrating several learnable Mona layers into the pre-trained model to align the training directions. Then, a label generator is designed to unify the reflection labels for both synthetic and real-world data. In addition, a Gaussian-based Adaptive Frequency Learning Block (G-AFLB) is proposed to adaptively learn and fuse the frequency priors, and a Dynamic Agent Attention (DAA) is employed as an alternative to window-based attention by dynamically modeling the significance levels across windows (inter-) and within an individual window (intra-). These components constitute our proposed Gap-Free Reflection Removal Network (GFRRN). Extensive experiments demonstrate the effectiveness of our GFRRN, achieving superior performance against state-of-the-art SIRR methods.
