You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Which one of those metric results are the ones you report in the paper?
I also noticed that when using eval_mode = 'epa', self.cfg_dist_ths and self.cfg_dist_th_tp are overwritten.
In the fourth metric calculation, which uses the MotionEval class initialized with category_convert_type='detection_category',
those thresholds are not reset: I was wondering if this in intentional.
Thank you.
The text was updated successfully, but these errors were encountered:
Hello,
In nuscenes_e2e_dataset.py, starting at line 1196, the motion metrics are calculated 6 times.
Those calculations are done in nuscenes_eval_motion.py :
eval_mode="standard":
bbox matching dist threshold: [0.5,1,2,4]
dist_th_tp = 2.0
Traj_dist_th = None
eval_mode="motion_map"
bbox matching dist threshold: [1.0]
dist_th_tp = 1.0
Traj_dist_th = 2.0
...
Which one of those metric results are the ones you report in the paper?
I also noticed that when using eval_mode = 'epa', self.cfg_dist_ths and self.cfg_dist_th_tp are overwritten.
In the fourth metric calculation, which uses the MotionEval class initialized with category_convert_type='detection_category',
those thresholds are not reset: I was wondering if this in intentional.
Thank you.
The text was updated successfully, but these errors were encountered: