Leaderboard Scene Parsing Car Instance Lane Segmentation Localization Trajectory Detection Tracking Stereo
CVPR Challenge rank 1
Megvii
0.3399
CVPR Challenge rank 2
Smart_Vision_SG
0.3022
CVPR Challenge rank 3
SZU_N606
0.2502
5
ChiZhouMeiziHao
0.24475
17
Azat Akhtyamov
0.18362
20
AlbertoCastaño
0.17734
23
xiaoyucoco1314
0.15463
Rank
Method
AP
AP50
AP75
APS
APM
APL
AR1
AR10
AR100
ARS
ARM
ARL
Rank
Method
WSADE
ADEv
ADEp
ADEb
WSFDE
FDEv
FDEp
FDEb
baseline
8.5881
7.9467
7.1811
12.8805
24.2262
12.7757
11.1210
22.7912
Team: Yuexin Ma, Xinge Zhu, Sibo Zhang, Ruigang
Yang, Wenping Wang, and Dinesh Manocha; BaiduRAL
Method Description:Based on
LSTM, our approach uses an instance layer to learn instances’ movements and
interactions and has a category layer to learn the similarities of instances
belonging to the same type to refine the
prediction.
Publication:
TrafficPredict: Trajectory Prediction for
Heterogeneous Traffic-Agents [
PDF]. AAAI(oral), 2019
Rank
Method
mAP
AP(Car)
AP(Pedestrian)
AP(Cyclist)
baseline
cnnseg
44.974
55.926
35.398
43.598
Rank
Method
MOTAm
MOTA(Car)
MOTA(Pedestrian)
MOTA(Cyclist)
baseline
IOU_tracker(based on cnnseg)
18.973
56.919
0
0
Rank
Method
D1_all
D1_fg
D1_bg