Skip to main content
. 2021 Dec 29;19(1):348. doi: 10.3390/ijerph19010348
Algorithm 1: Label generation for the dataset.
Input: The label of each frame of the dataset
Output: Label of each time window, where 0: other behaviors, 1: lane keeping, 2: acceleration, 3: deceleration, 4: turning, 5: lane change
For each time window:
if all frames in the time window are unlabeled then
Set 0 as the label of the time window
else
Compute the mode of the labels in the time window
Set the mode as the label of the time window
End for