The training data consists of 100 different sets. The sets are visualized below: the gray lines are the trajectories and the black line the corresponding ground truth road segment according to Open street map.
Input format:
Each set is in its own file.
Every file contains 3 or more trajectories.
Trajectories have been converted to UTM and normalized to the scale [0, 1].
Result format:
A single text file containing all representative trajectories.
Result should be coded in the same format as is the input.
Sample source code will be made available in all accepted languages.
You can test anytime how well your current method performs with the training data by dragging your results on this page and compare your result with others on the Rankings page.
