The Beautiful Future

tensorflow deeplab semantic segmentation 분석 본문

카테고리 없음

tensorflow deeplab semantic segmentation 분석

Small Octopus 2018. 5. 2. 14:14

싸이트

https://github.com/tensorflow/models/tree/master/research/deeplab


레퍼런스

[1] Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs, ICLR2015

[2] DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs, TPAMI2017

[3] Rethinking Atrous Convolution for Semantic Image Segmentation, arXiv

[4] Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation, arXiv


1. 사용된 기법

- atrous convolution을 사용함. [1]

- atrous convolution pyramid pooling(ASPP)을 사용함. [2]

- ASPP에 image-level feature를 augment함, batch normalization을 사용, 

 학습과 평가에서 다른 출력 스트라이드를 사용했다.  학습: 16, 평가: 8. [3]


Comments