Automatic Generation of Machine Learning Synthetic Data Using ROS

Data labeling is a time intensive process. As such, many data scientists use various tools to aid in the generation and While these help automate labeling, still require user interaction throughout Additionally, most target only few network frameworks. Any researchers exploring multiple frameworks must find additional orwrite conversion scripts. This paper presents an automated tool for generating synthetic arbitrary formats. It uses Robot Operating System (ROS) Gazebo, which are common robotics community. Through ROS paradigms, it allows extensive customization of simulation environment plugin-like framework development format writers without need change main body code. Using this tool, authors were able generate arbitrarily large image dataset three unique training formats using approximately 15 min setup variable amount hands-off run time, depending on size. The source code available at https://github.com/Navy-RISE-Lab/nn_data_collection