These are chat archives for beniz/deepdetect

21st
Dec 2016
Deepak Kumar Battini
@deepakkumar1984
Dec 21 2016 02:56
Hello All....
I am getting the following error while building deepdetect:
[ 33%] Built target caffe_dd
[ 91%] Built target ddetect
[ 95%] Linking CXX executable dede
/usr/bin/ld: warning: libprotobuf.so.11, needed by /home/deepak/deepdetect/build/caffe_dd/src/caffe_dd/build/lib/libcaffe.so, may conflict with libprotobuf.so.9
/usr/bin/ld: ../src/libddetect.a(caffelib.cc.o): undefined reference to symbol '_ZN6google8protobuf8internal26fixed_address_empty_stringB5cxx11E'
/usr/local/lib/libprotobuf.so.11: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
main/CMakeFiles/dede.dir/build.make:151: recipe for target 'main/dede' failed
make[2]: [main/dede] Error 1
CMakeFiles/Makefile2:177: recipe for target 'main/CMakeFiles/dede.dir/all' failed
make[1]:
[main/CMakeFiles/dede.dir/all] Error 2
Makefile:83: recipe for target 'all' failed
make: * [all] Error 2
deepak@deepak-ubuntu:~/deepdetect/build$
Any idea?
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 03:44
getting a weird error in docker
Error: image library/deepdetect_cpu:latest not found
any fix?
NVM, i'm a jing
you need to put /beniz/deepdetect_cpu
@beniz should probably update that in the documentation ;)
Idiots like me will have trouble otherwise D:
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 03:55
Also, I have a question about the image size. In the example, it has a size of 224x224, but the example image given is actually much larger. Does it resize the image?
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 04:49
got a weird error when trying to create the service
here's the code i'm using to create the service
curl -X PUT "http://localhost:8080/services/imageserv" -d "{\"mllib\":\"tensorflow\",\"description\":\"image classification service\",\"type\":\"supervised\",\"parameters\":{\"input\":{\"connector\":\"image\",\"height\":224,\"width\":224},\"mllib\":{\"nclasses\":1001,\"inputlayer\":\"InputImage\"}},\"model\":{\"repository\":\"/home/sixsamuraisoldier/Downloads/inception_resnet_v2.pb/\"}}"
and getting this back {"status":{"code":404,"msg":"NotFound","dd_code":1000,"dd_msg":"Unknown Library"}}
Emmanuel Benazera
@beniz
Dec 21 2016 08:33
@deepakkumar1984 you obviously have a conflict between two versions of the protonic library
Emmanuel Benazera
@beniz
Dec 21 2016 08:42
@sixsamuraisoldier you need a build with Tensorflow support. If you are using docker, the Tensorflow support is not built-in.
Emmanuel Benazera
@beniz
Dec 21 2016 08:55
for base64, data:["base64str",...]
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 16:47
@beniz ah thank you, that makes sense :)
@beniz are there plans to add tensorflow support to docker?
Emmanuel Benazera
@beniz
Dec 21 2016 16:52
@sixsamuraisoldier maybe... but not like immediately
There'll be an AMI with built-in TF support available in a few days, though I know this is not the same. We'll work on it after the holidays.
Note that the dockerfiles are part of the github repo, so if someone comes with a tested PR...
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 17:20
@beniz okay, that sounds good. I might work on it. Does it just require TF or does it have to be a special modified version for DD like Caffe did?
Emmanuel Benazera
@beniz
Dec 21 2016 17:31
Look at the README.md, and go with flag -DUSE_TF=ON. There'll be a bunch of questions when building TF, just use default answer, and choose whether you want GPU support.
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 17:35
@beniz TF's readme or DDs?
Emmanuel Benazera
@beniz
Dec 21 2016 17:36
DD
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 17:36
@beniz also is there a tutorial to use the caffe docker image and set up the server? The tutorial in the documentation assumes caffe is built from source it seems like
@beniz thanks
Emmanuel Benazera
@beniz
Dec 21 2016 17:36
look at the docker folder
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 17:37
excellent, thanks for bearing with me, I'm a total noob to DD xD
Emmanuel Benazera
@beniz
Dec 21 2016 17:37
np
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 17:52
@beniz Also, I see that input images are for 224x224 resolution but the example is a much larger picture, is it resized or cropped automatically?
@beniz Also, I can't seem to find the docker folder, could you point me to it by chance? Thx in advance
Emmanuel Benazera
@beniz
Dec 21 2016 18:06
the server resizes images yes
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 18:08
Awesome , thanks I was looking on the deepdetect.com website stupid me
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 18:30
so for setting up a service with caffe
on the docker image
do i just PUT the given request with the repository of the model pointing to the .caffemodel?
Is that the right file format?
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 18:59
How do I upload a model to the docker server?
Let's say I have my own model that I pretrained on imagenet
how am I supposed to upload it to the server?
Tapabrata Ghosh
@sixsamuraisoldier
Dec 21 2016 20:00
@beniz Is there any example for base64 encoding an image?
Emmanuel Benazera
@beniz
Dec 21 2016 20:36
For docker look at the documentation, and the share volume part
for base64, see my earlier comment, simply pass your base64 data strings