讀書筆記Fully Convolutional Networks for Semantic Segmentation

反捲積解釋 反捲積=逆卷積=上採樣=Deconvolution(caffe, keras)=conv_transpose(tensorflow)=nn.ConvTranspose2d(torch) 反捲積講的很好：https://blog.csdn.net/tfcy694/article/details/89073443 網絡分析 FCN網絡部分講的不錯：https://www.jianshu.c

>>阅读原文<<