arXiv:2606.26016v1 cs.CV

MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation

Yang Chen, Xiaowei Xu, Shuai Wang, Xinwen Zhang, Qiushi Guo, Tiezheng Ge, Limin Wang

제출 2026-06-24

원문 초록

Normalizing Flows (NFs) are powerful generative models capable of exact density estimation and sampling. However, their strict invertibility often forces the model to exhaust its capacity on low-level pixel details, hindering the capture of high-level semantic structures. While Masked Image Modeling (MIM) has excelled in representation learning, its integration into generative pipelines has remained largely modular and disjointed. In this paper, we propose MIMFlow, a unified end-to-end framework that jointly optimizes latent semantics, pixel reconstruction, and generative flow. By employing a VAE encoder to infer semantic latent from masked images, MIMFlow achieves a principled decoupling of the generative task: the Normalizing Flow focuses on modeling a simplified, low-frequency semantic manifold, while a specialized decoder handles high-frequency synthesis. This design effectively resolves the inherent capacity bottleneck of NFs, allowing the model to prioritize global structural coherence over redundant noise. Empirical results on ImageNet 256$\times$256 show that MIMFlow-L reaches 71.3\% linear probing accuracy and an FID of 2.50. Despite using only 128 tokens (50\% fewer than standard models), it yields a 32.8\% performance gain over similar-scale NF baselines. Our code is available at https://github.com/MCG-NJU/MIMFlow.

요약

Normalizing Flow(NF)의 고질적인 용량 병목 문제를 해결하기 위해, MIMFlow은 마스크 이미지 모델링(MIM)을 VAE 인코더와 NF에 엔드-투-엔드로 통합하여 의미론적 잠재 공간 모델링과 고주파 픽셀 합성을 원칙적으로 분리한다. ImageNet 256×256에서 FID 2.50, 선형 프로빙 정확도 71.3%를 달성하며, 128개의 토큰만으로 동규모 NF 기반 모델 대비 32.8% 성능 향상을 보인다.

제출 이력

확인된 버전만 표시합니다. 전체 제출 이력은 arXiv 원문에서 확인하세요.

  1. arXiv:2606.26016v1제출 시각 미확인

함께 살펴볼 논문

공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.

공통 태그로 연결된 공개 논문이 아직 없습니다.

토론

이 논문과 연결된 포럼 글이 아직 없습니다.