Green Detection

Translations: de

Detecting which pixels in the image belong to the playing field can make it easier and safer to detect other objects such as field lines, other robots and the ball. The field in RoboCup is green. So one way to recognize the field is to classify the pixels by their color, i.e. to recognize if they are green. However, this green detection depends strongly on the lighting and is also susceptible to shadows cast by the robots themselves.

The task of this seminar project is to compare the green detection of 4 different RoboCup SPL teams. For this purpose the algorithms be implemented in Python and evaluated afterwards. A data set will be provided for this purpose.

Optional Task:
Based on the comparison, an own solution can be developed. For example, with DNN's for semantic segmentation.

Relevant publications and theses:

Information about Nao Team Humboldt

RoboCup Regeln:
Public Github:
Internal Gitlab:

Source code of other RoboCup SPL teams

Public NaoTH Repo:
Public Nao Devils Repo:
Public B-human Repo:
rUNSWift: Hulks:

Publication lists of other teams

Possibly work of other teams can be helpful:

B-Human Abschlussarbeiten:
B-Human Publikationen:

More details

Supervision and technical support: Heinrich Mellmann Work in the lab necessary: possibly
Own PC necessary: yes