提出新型生成分割模型，可高效建模像素级高阶相关性。
Flow Stochastic Segmentation Networks
- 采用离散与连续时间流架构，突破传统低秩参数限制
- 无需假设秩或存储参数，可估计任意高秩像素协方差
- 采样效率高于扩散模型，适合医学图像分割任务
我们提出流动随机分割网络（Flow Stochastic Segmentation Network, Flow-SSN），一类包含离散时间自回归和现代连续时间流变体的生成分割模型。我们证明了以往方法中低秩参数化的根本局限，并表明Flow-SSNs可在不假设秩或存储分布参数的情况下，估计任意高秩的像素级协方差。由于模型大部分容量用于学习流动的基础分布，形成一个表达性强的先验，因此其采样效率高于标准扩散模型。我们在具有挑战性的医学影像基准测试中应用Flow-SSNs，取得了当前最优结果。代码已开源：https://github.com/biomedia-mira/flow-ssn。
原文摘要 · Abstract (English)
We introduce the Flow Stochastic Segmentation Network (Flow-SSN), a generative segmentation model family featuring discrete-time autoregressive and modern continuous-time flow variants. We prove fundamental limitations of the low-rank parameterisation of previous methods and show that Flow-SSNs can estimate arbitrarily high-rank pixel-wise covariances without assuming the rank or storing the distributional parameters. Flow-SSNs are also more efficient to sample from than standard diffusion-based segmentation models, thanks to most of the model capacity being allocated to learning the base distribution of the flow, constituting an expressive prior. We apply Flow-SSNs to challenging medical imaging benchmarks and achieve state-of-the-art results. Code available: https://github.com/biomedia-mira/flow-ssn.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。