Contains Datasets that describe the relation of flags seen by a player to its position at the field and the direction it is currently looking at.
Position is relative to the Side the player is currently playing.
Value ranges:
- x: -50 to 50
- y: -30 to 30
- direction: -180 to 180
x, y values outside the range are possible but lay outside the field.
Position and direction seen from the east and west side.
The features contained in the dataset define weather a player sees a certain flag at his current position and the direction he is looking. There is one value for each of the 41 flags the player can see. The value is 0 if the player does not see the flag. If the value is greater than 0, the player can see the flag. The value is the angle + 100 in which the player sees the flag.
The following image explains the feature values for one player and three flags.
- Flag A: The player sees the flag in an angle of +10° → The feature value is 100 + 10 = 110.
- Flag B: The player sees the flag in an angle of -25° → The feature value is 100 – 25 = 75.
- Flag C: The player does not see the flag → The feature value is 0.
Flags seen from east and west side
- 0: Rownumber
- 1: X-Position
- 2: Y-Position
- 3: Direction
- 4 - 45: Features for Flags 0 - 41
Prerequisites
- Have a jdk installed on your comuter
- Have sbt installed on your comuter
- Have git installed on your computer
- Have maven installed on your computer
Steps
- clone the vsoc repository. call 'git clone https://github.com/wwagner4/vsoc.git' to get the latests source code
- change to the directory vsoc/vsoc-2007/ and call 'mvn install' to compile the old java code that is needed to create new data
- change to the directory vsoc/machinelearning-2017/ and call 'sbt ";project create_data;runMain playerpos.PlayerposCreateDataMain"' to create new datasets