This page has the most up-to-date information for our challenges. For detailed information on a method, please click the method name. To sort by a specific metric, click on the header in the table. For further questions, please contact us at

Additional Information Used

  • Individual Image: Method uses individual images from each camera
  • Stitched Image: Method uses stitched images combined from the individual cameras
  • Pointcloud: Method uses 3D pointcloud data
  • Online Tracking: Method does frame-by-frame processing with no lookahead
  • Offline Tracking: Method does not do in-order frame processing
  • Public Detections: Method uses publicly available detections
  • Private Detections: Method uses its own private detections

2D Detection Leaderboard

Name AP ↑ Runtime ↓ CPU/GPU
Faster R-CNN
52.17 0.038 s 1 GPU (Titan X)
S. Ren, K. He, R. Girshick and J. Sun. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. In NeurIPS, 2015.
50.38 0.056 s 1 GPU (Titan X)
T. Lin, P. Goyal, R. Girshick, K. He and P. Dollár. Focal Loss for Dense Object Detection. In ICCV, 2017.
48.51 0.35 s 1 GPU (GTX 1060)
Anonymous Submission
41.73 0.051 s 1 GPU (Titan X)
J. Redmon and A. Farhadi. YOLOv3: An Incremental Improvement. In arXiv, 2018.

3D Detection Leaderboard

Name AP ↑ Runtime ↓ CPU/GPU
72.70 1.0 s GTX 1080 Ti
Anonymous Submission
63.92 0.28 s 1 GPU (Tesla K40c)
Anonymous Submission
54.94 0.28 s 1 GPU (Titan Tesla K40c)
Zhe Liu, Xin Zhao, Tengteng Huang, Ruolan Hu, Yu Zhou and Xiang Bai. TANet: Robust 3D Object Detection from Point Clouds with Triple Attention. In AAAI, 2020.
42.78 0.019 s 1 GPU(TITAN V)
Anonymous Submission
38.21 0.17 s 1 GPU (Titan X)
C. Qi, W. Liu, C. Wu, H. Su and L. Guibas. Frustum PointNets for 3D Object Detection from RGB-D Data. In CVPR, 2018.