This repository contains the source code for the semantic image segmentation method described in the ICCV 2015 paper: Conditional Random Fields as Recurrent Neural Networks. http://crfasrnn.torr.vision/
I'm training CRFasRNN layer on top of upsampled score_final layer similar to that in FCN8s on my data, and I quickly noticed that the output from CRF-RNN is not very different to score_final. What could be the reason for that? Just the learning rate or something more interesting?
I'm training CRFasRNN layer on top of upsampled score_final layer similar to that in FCN8s on my data, and I quickly noticed that the output from CRF-RNN is not very different to score_final. What could be the reason for that? Just the learning rate or something more interesting?