YujunXie / Papers-of-Crowd-Counting

0 stars 1 forks source link

HRNet #1

Open YujunXie opened 4 years ago

YujunXie commented 4 years ago

HRNet

《Deep High-Resolution Representation Learning for Human Pose Estimation》
  1. 简介
  2. 方法
  3. 实验
  4. 思考
  5. 参考文献

简介

方法

image-20191127204518416

It generates reliable high-resolution representations through repeatedly fusing the representations produced by the high-to-low subnetworks.

:arrow_up::最近邻上采样 + 1x1conv;:arrow_down::3x3conv, stride=2; :arrow_right:: 平移复制;

不同尺度特征图融合:concat;

重复使用低分辨率的特征图增强高分辨率表示。最后得到的特征图既能保持高分辨表征,还具有多尺度信息。

  1. parallel high-to-low resolution subnetworks
  2. multi-resolution subnetworks (multi-scale fusion)

实验

image-20191128155157958

网络输入图像尺寸的影响:HRNet在小尺寸128x96图像上的提升最为显著,能够证明高分辨率表征对小目标的作用。

image-20191128155701671

思考:thinking:

:question:文章中关于exchange unit对应multi-scale fusion的叙述让人摸不着头脑

:dart: take-home-message

:sparkles: 网络结构简单高效,可根据训练目标进行调整:

:sparkles: 多尺度信息的融合方式可以更加丰富和可解释性,如semantic-guided, weight-based.

参考文献:books:

论文阅读HRNetV1,HRNetV2,HRNetV2p

《Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs》

《Multi-scale structure-aware network for human pose estimation》

《Fast multi-person pose estimation using pose residual network》

《Pose partition networks for multi-person pose estimation》

《Pose proposal networks》

《Integral human pose regression》

《Deeply learned compositional models for human pose estimation》

YujunXie commented 4 years ago

HigherHRNet

《Bottom-Up Higher-Resolution Networks for Multi-Person Pose Estimation》

简介

方法

image-20191129104710479

the input to our deconvolution module is the concatenation of the feature maps and the predicted heatmaps from either HRNet or previous deconvolution modules.

在经过HRNet后的特征图经过4x4的反卷积上采样2倍,再通过4个residual block输出heatmap。

实验

思考:thinking:

:question:创新点

:dart: take-home-message

:sparkles: 网络拼接工作,美文一篇,实验做的详实,可以作为写论文的参考(从提出问题,解决问题,实验设计等)。

参考文献:books:

《Searching for efficient multi-scale ar- chitectures for dense image prediction》