Libor Wagner

I had same problem, so I downgraded to keras=1.2.2 then saved entire model with `model.save_model`, then upgrade to keras=2.0.4 load the model and it works fine...with some warning to migrate...

The idea is that i have Pweave document (script), which generates report, now I would like to generate same report for different input data, ideally without the need to change...

Yes that would be great.

We have already implemented an IO interface for motoman (DX100) robot. With bit of help we can make it more general. The current implementation uses another simple message. The message...

@gavanderhoorn: yes, the motoman side is [here](http://clopema.felk.cvut.cz/gitweb/?p=ros_industrial.git;a=tree;f=motoman/dx100/motoplus/motoros_lib;h=28bed42d0d9d3eee28a140097a51afb9a79fb7b5;hb=HEAD) io_interface.[cpp,h], and the message is [here](http://clopema.felk.cvut.cz/gitweb/?p=ros_industrial.git;a=tree;f=industrial_core/simple_message;h=4f34c2d9e9d6414fd11d2dd76e9e67aa0099ad22;hb=HEAD) io_data.[cpp,h]

Actually it is a ROS stuff, the driver reads the `PS1080.ini` but then the configuration in overwritten by the hard coded pixel format in the https://github.com/ros-drivers/openni2_camera/blob/indigo-devel/src/openni2_driver.cpp#L179. Is this intentional?