Search papers, labs, and topics across Lattice.
This paper introduces Spectrum-Adaptive Scheduling (SAS) and Measurement-Prioritized Attention (MPA) as complementary components to enhance flow-based generative models for image inverse problems. By optimizing the allocation of function evaluations over time and guiding spatial information flow, the methods effectively balance early exploration and late-stage refinement, addressing critical shortcomings in existing solvers. Experimental results across various inverse problems, including super-resolution and motion deblurring, show significant improvements in restoration quality without the need for retraining or additional evaluations.
Optimizing function evaluations in flow-based generative models can dramatically enhance image restoration quality without retraining.
Flow-based generative models have emerged as powerful image priors for training-free inverse problem solving, capturing coherent semantics and fine-grained structure. Despite these strengths, existing flow-based inverse solvers primarily focus on the design of individual updates, largely overlooking spatio-temporal information allocation under a fixed number of function evaluations (NFEs). Temporally, insufficient early exploration can trap the flow trajectory in an incorrect semantic basin, whereas excessive allocation of NFEs to early stages leaves little budget for late-stage refinement. Spatially, data consistency provides direct constraints only within observed regions, whereas the recovery of missing regions relies mainly on the generative prior. To address these two issues, we introduce two complementary and training-free components, i.e., Spectrum-Adaptive Scheduling (SAS) and Measurement-Prioritized Attention (MPA). For temporal allocation, SAS distributes the available NFEs over flow time according to the degradation spectrum and logSNR geometry, thus better balancing semantic exploration and detail refinement. For spatial propagation, MPA exploits data-prior conflicts to guide information toward weakly constrained regions, thereby enhancing semantic and structural fidelity. Extensive experiments on standard image inverse problems, e.g., super-resolution, motion deblurring, and inpainting, demonstrate that the proposed components can be integrated into existing flow-based inverse solvers in a plug-and-play manner without retraining or additional flow-model evaluations, and can also significantly improve the restoration quality of existing solvers.
