richzhang / colorization

Automatic colorization using deep neural networks. "Colorful Image Colorization." In ECCV, 2016.
http://richzhang.github.io/colorization/
BSD 2-Clause "Simplified" License
3.34k stars 923 forks source link

./train/fetch_caffe.sh gets different version of Caffe #40

Open leekkww opened 6 years ago

leekkww commented 6 years ago

It seems that ./train/fetch_caffe.sh doesn't get the modified version of Caffe that it is supposed get. The downloaded tar file opens to to a directory called caffe_private_pascal/ (instead of caffe-colorization/, which is what is called by ./train/train_model.sh), and the presets for Makefile.config.example does not match with what the model asks for. I had to change a bunch of things after Googling really hard and comment/uncomment a bunch of other things to get the model to to work. Did you remove the supported version of Caffe?

yuanyang7 commented 6 years ago

I got the same problem.😖 How to solve it..

chnagi commented 6 years ago

I too got the same problem. Did Anyone of you resolved this issue? Please let me know

xuyifangreeneyes commented 6 years ago

I also got the same problem. QAQ

alwyz commented 5 years ago

Check here

https://people.eecs.berkeley.edu/~rich.zhang/projects/2016_colorization/files/train/

EDIT: I tried the link above and indeed it seems to point to the original referenced version of caffe-colorization, but no matter what version of cudnn i have installed on my system, it seems to crash during the make / maketest / runtest process...

alwyz commented 5 years ago

OK, so....

I successfully downloaded and built caffe-colorization with this method on a new server with python 2.7 pre-installed. I am actually training for the first time!

I have cuda 8.0 installed . Cudnn v4 (tho i dont think cudnn matters, since i disable it in the makefile)

https://chunml.github.io/ChunML.github.io/project/Installing-Caffe-Ubuntu/

Stop following install info on that site when it gets to the part about installing caffe.

Do the following instead

git clone -b master --single-branch https://github.com/richzhang/colorization.git

run ./train/fetch_init_model.sh

Then, modify /train/fetch_caffe.sh file to use this file in wget line:

https://people.eecs.berkeley.edu/~rich.zhang/projects/2016_colorization/files/train/caffe-colorization_old.tar.gz

run ./train/fetch_caffe.sh

then, in caffe-colorization directory:

modify makefile.config with the following

Comment out cudnn. Couldnt get it to work.

Change CV settings to comment the current one, and uncomment the #3 one several lines below

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial/

comment out the matlab line (i dont have matlab so...)

uncomment python layer:=1

then

make all & make test && make runtest && make pycaffe

dont forget to add your path to your new caffe to bashrc

sudo vim ~/.bashrc

export PYTHONPATH=$HOME/colorization/colorization/caffe-colorization/python:$PYTHONPATH export PYTHONPATH=$HOME/colorization/colorization/resources:$PYTHONPATH

and then run

source ~/.bashrc

There's still a few post install bugs that are googleable. But this got me going in the right direction. Hope it helps someone else.