The KITTI Vision Benchmark Suite

The impact of occlusion scenarios of the objects
on the keypoints prediction is often overlooked.
To address this issue, we proposes a novel 3D
monocular object detection method named MonOAPC,
which is equipped with occlusion-aware plane-
constraints.

learning_rate = 2e − 4,
weight_decay = 1e − 5,

@article{yao2023occlusion,
title={Occlusion-Aware Plane-Constraints for
Monocular 3D Object Detection},
author={Yao, Hongdou and Chen, Jun and Wang,
Zheng and Wang, Xiao and Han, Pengfei and Chai,
Xiaoyu and Qiu, Yansheng},
journal={IEEE Transactions on Intelligent
Transportation Systems},
year={2023},
publisher={IEEE}
}