论文阅读 [TPAMI-2022] Index Networks

论文阅读 [TPAMI-2022] Index Networks

论文搜索(studyai.com)

搜索论文: Index Networks

搜索论文: http://www.studyai.com/search/whole-site/?q=Index+Networks

关键字(Keywords)

Indexes; Task analysis; Interpolation; Semantics; Image segmentation; Convolution; Estimation; Upsampling operators; dynamic networks; image denoising; semantic segmentation; image matting; depth estimation

机器学习; 机器视觉

检测分割; 编码解码器; 图像/视频去噪声; 图像抠图; 深度估计

摘要(Abstract)

We show that existing upsampling operators in convolutional networks can be unified using the notion of the index function.

我们证明了卷积网络中现有的上采样算子可以用索引函数的概念统一起来。.

This notion is inspired by an observation in the decoding process of deep image matting where indices-guided unpooling can often recover boundary details considerably better than other upsampling operators such as bilinear interpolation.

这一概念的灵感来源于深度图像抠图解码过程中的一个观察，其中索引引导的去冷却通常可以比其他上采样算子（如双线性插值）更好地恢复边界细节。.

By viewing the indices as a function of the feature map, we introduce the concept of ‘learning to index’, and present a novel index-guided encoder-decoder framework where indices are learned adaptively from data and are used to guide downsampling and upsampling stages, without extra training supervision.

通过将索引视为特征映射的函数，我们引入了“学习索引”的概念，并提出了一种新的索引引导编码器-解码器框架，其中索引从数据中自适应学习，并用于指导下采样和上采样阶段，无需额外的训练监督。.

At the core of this framework is a new learnable module, termed Index Network (IndexNet), which dynamically generates indices conditioned on the feature map.

该框架的核心是一个新的可学习模块，称为索引网络（IndexNet），它根据特征图动态生成索引。.

IndexNet can be used as a plug-in, applicable to almost all convolutional networks that have coupled downsampling and upsampling stages, enabling the networks to dynamically capture variations of local patterns.

IndexNet可以用作插件，适用于几乎所有具有耦合下采样和上采样级的卷积网络，使网络能够动态捕获局部模式的变化。.

In particular, we instantiate and investigate five families of IndexNet.

特别是，我们实例化并研究了IndexNet的五个家族。.

We highlight their superiority in delivering spatial information over other upsampling operators with experiments on synthetic data, and demonstrate their effectiveness on four dense prediction tasks, including image matting, image denoising, semantic segmentation, and monocular depth estimation.

通过对合成数据的实验，我们强调了它们在传递空间信息方面优于其他上采样算子，并证明了它们在四项密集预测任务上的有效性，包括图像去噪、语义分割和单目深度估计。.

Code and models are available at https://git.io/IndexNet…

代码和模型可在https://git.io/IndexNet…

作者(Authors)

[‘Hao Lu’, ‘Yutong Dai’, ‘Chunhua Shen’, ‘Songcen Xu’]