论文阅读理解 - Semantic Image Segmentation With Deep Convolutional Nets and Fully Connected CRFs
2017-10-12 16:47
1551 查看
Semantic Image Segmentation With Deep Convolutional Nets and Fully Connected CRFs
Project摘要 —— 主要是将CNN和概率图模型结合,来处理像素级分类问题,即语义图像分割. 由于CNN具有不变性,适合于 high-level 任务,如图像分类. 但CNN网络最后一层的输出不足以精确物体分割. 这里结合CNN最后输出层的特征与全连接CRF相结合,提升语义分割效果.
语义分割的目标——Goal:
将图像分割为不同的语义部分,并对不同部分进行分类.
主要是两部分:
采用CNN来初步得到图像分割结果
hole algorithm (atrous algorithm),应对subsampling问题.
train 阶段 - loss function 采用 CNN输出特征图的各像素空间位置的 sum of cross-entropy term (SoftmaxWithLoss). 原始图片的尺寸是CNN输出特征图的 8 倍, ground truth labels 也降采样到与CNN输出尺寸一致.
test 阶段 - 将CNN输出转换到原始图像分辨率. CNN输出图具有很好的平滑性,故采用 bilinear interpolation 来增加其分辨率(factor=8).
采用CRF将CNN分割结果精细化
spatial invariance
1. CNN
2. CRF
3. Results
好的分割结果:
差的分割结果:
4. Conclusion
Modify the CNN architecture to become less spatially invariant.Use the CNN to compute a rough score map.
Use a fully connected CRF to sharpen the score map.
5. Reference
[1] - Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs[2] - Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs - slides
[3] - 深度学习轻松学-核心算法与视觉实践 - 第九章-应用:图像的语意分割
相关文章推荐
- 深度学习论文(八)---DeepLabV1-SEMANTIC IMAGE SEGMENTATION WITH DEEP CONVOLUTIONAL NETS AND FULLY CONNECTED C
- 《Semantic image segmentation with deep convolution nets and fully connected CRFs》学习总结(deeplabV1)
- 【deeplab】Semantic Image Segmentation with Deep Convolutional Nets and Fully
- Semantic Segmentation -- (DeepLabv1)Semantic image segmentation with deep convolutional ... CRFs论文解读
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution--阅读笔记
- 深度学习论文(九)---DeepLabV2-Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution,
- 论文阅读(1)——ImageNet Classification with Deep Convolutional Neural Networks
- SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation 论文阅读
- ImageNet Classification with Deep Convolutional Neural Networks 阅读理解及问题
- AlexNet论文--ImageNet Classification with Deep Convolutional Neural Networks+个人理解
- 论文阅读理解 - Fully Convolutional Instance-aware Semantic Segmentation
- 论文阅读ImageNet Classification with Deep Convolutional Neural Networks &Going Deeper with Convolutions
- Semantic Segmentation -- (DeepLabv2)Semantic Image Segmentation ... Fully Connected CRFs论文解读
- Fully Convolutional Networks for Semantic Segmentation论文阅读
- Deep learning论文笔记一:ImageNet Classification with Deep Convolutional Neural Networks
- DeepLab:深度卷积网络,多孔卷积 和全连接条件随机场 的图像语义分割 Semantic Image Segmentation with Deep Convolutional Nets, Atro
- 论文阅读笔记:Fully Convolutional Networks for Semantic Segmentation
- 论文阅读理解 - (Deeplab-V3)Rethinking Atrous Convolution for Semantic Image Segmentation
- ImageNet Classification with Deep Convolutional Neural Networks 论文解读
- [深度学习论文笔记][Image Classification] ImageNet Classification with Deep Convolutional Neural Networks