In this episode I discuss the paper "Fully Convolutional Networks for Semantic Segmentation" in detail. This work as been a big-breakthrough as it proposed a better way to approach the semantic segmentation task with CNNs and overcome some important limitations.
### Links ###
Paper:
https://people.eecs.berkeley.edu/~jonlong/long_shelhamer_fcn.pdf
Presentation-pdf:
www.micc.unifi.it/bagdanov/pdfs/FCN-presentation.pdf
Code Caffe:
https://github.com/shelhamer/fcn.berkeleyvision.org
Code Tensorflow;
https://github.com/MarvinTeichmann/tensorflow-fcn
