I am trying to use this repository for tracking with my own model detecting custom classes. My model outputs the boxes, scores and labels in the same format as yours.
I do not have access to any oxts file.
It appears that you have partially accounted for this in tracking.py line 798:
if Path(oxts_file).exists(): # ego motion compensation will be applied if oxts data is given vf,vl,vz = load_oxts(oxts_file) mot_trackers = {tracking_name: MOT(tracking_name=tracking_name, calib=calib,oxts=(vf,vl,vz)) for tracking_name in TRACKING_CLASSES} else: mot_trackers = {tracking_name: MOT(tracking_name=tracking_name, calib=calib,oxts=None) for tracking_name in TRACKING_CLASSES}
However in the KalmanBoxTracker class, the update() function requires oxts in line 292:
vf = oxts[0][frame-1] vl = oxts[1][frame-1] vz = oxts[2][frame-1]
Which causes a crash when attempting to run the tracking algorithm.
Is there a way to get around this? Should I set the oxts to 0? Any help would be appreciated.
I am trying to use this repository for tracking with my own model detecting custom classes. My model outputs the boxes, scores and labels in the same format as yours.
I do not have access to any oxts file.
It appears that you have partially accounted for this in tracking.py line 798:
if Path(oxts_file).exists(): # ego motion compensation will be applied if oxts data is given vf,vl,vz = load_oxts(oxts_file) mot_trackers = {tracking_name: MOT(tracking_name=tracking_name, calib=calib,oxts=(vf,vl,vz)) for tracking_name in TRACKING_CLASSES} else: mot_trackers = {tracking_name: MOT(tracking_name=tracking_name, calib=calib,oxts=None) for tracking_name in TRACKING_CLASSES}
However in the KalmanBoxTracker class, the update() function requires oxts in line 292:
vf = oxts[0][frame-1] vl = oxts[1][frame-1] vz = oxts[2][frame-1]
Which causes a crash when attempting to run the tracking algorithm.
Is there a way to get around this? Should I set the oxts to 0? Any help would be appreciated.