- Fully Convolutional Networks
summary (easy to understand): 卷积神经网络CNN（3）—— FCN(Fully Convolutional Networks)要点解
- important ideas:
- upsampling
- end to end - softmax pixel wise classification
original paper: Fully Convolutional Networks for Semantic Segmentation
- UNet: built based on FCN
()[https://arxiv.org/pdf/1505.04597.pdf)%E5%92%8C%5bTiramisu%5d(https://arxiv.org/abs/1611.09326.pdf]
- Mask RCNN
- Mattport Mask RCNN package
- Kaiming He Paper
