[深度学习论文笔记][Image Reconstruction] Understanding Deep Image Representations by Inverting Them

最新推荐文章于 2024-03-06 16:32:57 发布

Hao_Zhang_Vision

最新推荐文章于 2024-03-06 16:32:57 发布

阅读量3.2k

点赞数

CC 4.0 BY-SA版权

分类专栏： CNN Papers 文章标签： CNN Deep Learning Papers Computer Vision Image Reconstruction

本文链接：https://blog.youkuaiyun.com/Hao_Zhang_Vision/article/details/52980048

本文探讨了深度学习中图像表示的理解，通过逆向重构来揭示CNN如何编码图像。研究发现，表示会忽略如光照或视角等不相关差异。实验使用了p-norm和总变差（TV）作为正则化器，平衡不同损失项以得到重构图像。结果表明，卷积层保持了图像的真实感，全连接层则重构为类似但不完全相同的图像组成部分，丢失位置信息但保留了某些区分特征。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Mahendran, Aravindh, and Andrea Vedaldi. “Understanding deep image representations by inverting them.” 2015 IEEE conference on computer vision and pattern recognition

(CVPR). IEEE, 2015. (Citations: 142).

1 Motivation
Given an intermediate layer’s representation of an image, to which extent is it possible to reconstruct the image itself?

Note that the representations collapse irrelevant differences in images (e.g. illumination or viewpoint), so that the representation should not be uniquely invertible.

2 Idea

Starting from random noise, find an image whose representations best matches the given one.