The KITTI Vision Benchmark Suite

Monocular 3D object detection using
complementary depths.
The model predicts multiple depth estimates from
different
geometric representations and combines them
through a
learned depth ensemble. Built on CenterNet-
based architecture
with DLA-34 backbone. Trained on KITTI training
split.

None

@inproceedings{yan2024monocd,
title={MonoCD: Monocular 3D Object Detection
with Complementary Depths},
author={Yan, Longfei and Yan, Pei and Xiong,
Shenghua and Xiang, Xuanyu and Tan, Yihua},
booktitle={CVPR},
year={2024}
}