Untangling Local and Global Deformations in Deep Convolutional Networks for Image Classification and Sliding Window Detection
George Papandreou Toyota Technological Institute at Chicago Iasonas Kokkinos and Pierre-André Savalle Ecole Centrale Paris and INRIA
Abstract
Deep Convolutional Neural Networks (DCNNs) commonly use generic ‘max-pooling’ (MP) layers to extract deformation-invariant features, but we argue in favor of a more refined treatment. First, we introduce epitomic convolution as a building block alternative to the common convolution-MP cascade of DCNNs; while having identical complexity to MP, Epitomic Convolution allows for parameter sharing across different filters, resulting in faster convergence and better generalization. Second, we introduce a Multiple Instance Learning approach to explicitly accommodate global translation and scaling when training a DCNN exclusively with class labels. For this we rely on a ‘patchwork’ data structure that efficiently lays out all image scales and positions as candidates to a DCNN. Factoring global and local deformations allows a DCNN to ‘focus its resources’ on the treatment of non-rigid deformations and yields a substantial classification accuracy improvement. Third, further pursuing this idea, we develop an efficient DCNN sliding window object detector that employs explicit search over position, scale, and aspect ratio. We provide competitive image classification and localization results on t
中文速览
深度卷积神经网络在处理图像变形（平移、缩放、非刚性形变）时通常依赖通用的最大池化（max-pooling）层，但这种做法较为粗糙。本文提出了三项改进：一是用"epitomic convolution（外延卷积）"替代传统卷积加最大池化的组合，通过让相近滤波器共享参数来降低模型冗余，在计算量不变的前提下实现更快收敛和更好泛化；二是引入多示例学习（Multiple Instance Learning）框架，借助一种叫"patchwork"的数据结构在训练和测试时高效枚举图像的不同位置与尺度，让网络显式处理全局平移和缩放，从而把有限容量集中用于应对局部非刚性形变；三是在此基础上构建了一个滑动窗口目标检测器，对位置、尺度和长宽比进行显式搜索。在ImageNet ILSVRC-2012分类任务上，所提方法以单网络实现了10.0%的top-5错误率，在Pascal VOC 2007检测基准上也取得了有竞争力的结果，证明显式建模全局变形与引入参数共享的上位卷积层可以切实提升大规模视觉识别的性能。
原文 arXiv:1412.0296;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1412.0296v1