search

torrvision / crfasrnn

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/
Other
1.34k stars 460 forks source link

Training with my own data : Multiple problems arising #152

Open matthiasmace opened 6 years ago

matthiasmace commented 6 years ago

Hello I am trying to train cfrasrnn with my own data using https://github.com/martinkersner/train-CRF-RNN/ I had to slightly modify some pieces of the code to make it run on my Mac (OSX.12) but it now runs the training. I have one first problem when trying to parse the log file in order to vizualise loss (using loss_from_log.py https://github.com/martinkersner/train-CRF-RNN/blob/master/loss_from_log.py) : _Traceback (most recent call last): File "loss_from_log.py", line 123, in main() File "loss_from_log.py", line 43, in main trainloss.append(float(matched.group(1))) AttributeError: 'NoneType' object has no attribute 'group'

Indeed, in the log file, there's no "loss-ft" output. Why ?

The second is that when I try to use the caffemodel output (e.g. iteration 5000 or 7000) to test if it recognizes object, it now fails to do so (I tried with various objects). What am I doing wrong please ? Many thanks if any one can help !

Sam813 commented 6 years ago

@matthiasmace I have the same problem with loss_from_log. Have you solved it?