zlarumbe / roboflex_visualization Goto Github PK

View Code? Open in Web Editor NEW

Roboflex support for visualizing rgb and depth streams using SDL2.

License: MIT License

C++ 59.89% Python 17.67% CMake 22.44%

roboflex_visualization's Introduction

roboflex.visualization

A thin wrapper of various display nodes over SDL2. For now just supplies one node: RGBImageTV. It can display rgb tensors of the form (H,W,3) uint8.

System requirements

libsdl: SDL (Simple Directmedia Layer)

  # search for existing libsdl packages
  dpkg -l | grep sdl # -> shows nothing

  # install libsdl
  sudo apt-get install libsdl2-dev

  # search for existing libsdl packages again
  dpkg -l | grep sdl # -> shows:
  ii  libsdl2-2.0-0:amd64                        2.0.20+dfsg-2ubuntu1.22.04.1                                        amd64        Simple DirectMedia Layer
  ii  libsdl2-dev:amd64                          2.0.20+dfsg-2ubuntu1.22.04.1                                        amd64        Simple DirectMedia Layer development files

  THIS, for me on ubuntu22, puts headers in /usr/include/SDL2 and libs in /lib/x86_64-linux-gnu/libSDL2-2.0.so

  # to remove it:
  sudo apt-get -y purge libsdl2-dev

Import

import roboflex.visualization as rv

Nodes

There are two:

There is are two: RGBImageTV an DepthTV

# RGBImageTV will redraw rgb images at some
# frequency (refresh rate). It will look into
# an messages it receives - those that contain
# the key "rgb", mapping to a (3,H,W) tensor
# will be rendered. Messages get propagated.

rgb_tv = rcw.RGBImageTV(

    # mandatory (refresh rate)
    frequency_hz: float,

    # optional; defaults are:
    width = 640,
    height = 480,
    image_key = "rgb",
    initial_pos = (-1, -1),
    mirror = False,
    debug = False,
    name = "RGBImageTV"
)

# must be started!
rgb_tv.start()


# RGBImageTV will redraw rgb images at some
# frequency (refresh rate). It will look into
# an messages it receives - those that contain
# the key "depth", mapping to a (H,W) tensor
# will be rendered. Messages get propagated.

depth_tv = rcw.DepthTV(

    # mandatory (refresh rate)
    frequency_hz: float,

    # optional; defaults are:
    width = 640,
    height = 480,
    image_key = "depth",
    initial_pos = (-1, -1),
    mirror = False,
    debug = False,
    name = "RGBImageTV"
)

# must be started!
depth_tv.start()

Recommend Projects