基于图神经网络的自适应学习式图像压缩
Adaptive Learned Image Compression with Graph Neural Networks
用双尺度图结构替代 CNN 固定核与窗口注意力，按局部内容复杂度动态调整每个节点的邻居数，为视觉任务提供一种数据驱动的自适应感受野建模手段，可直接迁移到需要捕捉长程冗余的复原与压缩网络设计。
原文依据 we propose a content-adaptive image compression framework based on Graph Neural Networks (GNNs)
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论高效的图像压缩依赖于对局部与全局冗余的建模。目前最优的学习式图像压缩（LIC）方法大多基于 CNN 或 Transformer，其结构本质上是刚性的。标准 CNN 卷积核与基于窗口的注意力机制施加了固定的感受野与静态的连接模式，这可能仅仅因为像素在欧氏空间中相邻，就把非冗余的像素耦合在一起。这种刚性限制了模型自适应地捕捉图像中随空间变化的冗余的能力，尤其是在全局层面。为克服这些局限，我们提出了一种基于图神经网络（GNN）的内容自适应图像压缩框架。具体而言，我们的方法构建双尺度图，从而实现灵活、数据驱动的感受野。此外，我们引入自适应连接，根据局部内容复杂度动态调整每个节点的邻居数量。这些创新使我们的基于图的学习式图像压缩（GLIC）模型能够有效建模图像中多样的冗余模式，从而实现更高效、更具自适应的压缩。实验表明，GLIC 达到了最优性能，在 Kodak、Tecnick 和 CLIC 上相对于 VTM-9.1 分别取得 19.29%、21.69% 和 18.71% 的 BD-rate 降低。代码将发布于 https://github.com/UnoC-727/GLIC。
英文原文摘要 1408 字符
Efficient image compression relies on modeling both local and global redundancy. Most state-of-the-art (SOTA) learned image compression (LIC) methods are based on CNNs or Transformers, which are inherently rigid. Standard CNN kernels and window-based attention mechanisms impose fixed receptive fields and static connectivity patterns, which potentially couple non-redundant pixels simply due to their proximity in Euclidean space. This rigidity limits the model's ability to adaptively capture spatially varying redundancy across the image, particularly at the global level. To overcome these limitations, we propose a content-adaptive image compression framework based on Graph Neural Networks (GNNs). Specifically, our approach constructs dual-scale graphs that enable flexible, data-driven receptive fields. Furthermore, we introduce adaptive connectivity by dynamically adjusting the number of neighbors for each node based on local content complexity. These innovations empower our Graph-based Learned Image Compression (GLIC) model to effectively model diverse redundancy patterns across images, leading to more efficient and adaptive compression. Experiments demonstrate that GLIC achieves state-of-the-art performance, achieving BD-rate reductions of 19.29%, 21.69%, and 18.71% relative to VTM-9.1 on Kodak, Tecnick, and CLIC, respectively. Code will be released at https://github.com/UnoC-727/GLIC.