ROS nodes perform inference on a hard coded image

[gmin7]

Description:

• Here, the nodes you've setup will actually read in hardcoded images and detect cones in them

Goals:

• After this issue is completed, we can load in training set images, and see the bounding boxes of them printed out

*Acceptance Criteria:

• Can load in a hardcoded png file that the ROS nodes perform inference on
• The cone detection publisher node actually prints out bounding boxes of what it sees (test on the training set to make sure it's totally working)