Method

Attention-based Depth Distillation for Monocular 3D Object Detection [ADD]
https://arxiv.org/abs/2211.16779

Submitted on 8 Aug. 2022 19:03 by
Yunzhe Wu (University of Cambridge)

Running time:0.1 s
Environment:1 core @ 2.5 Ghz (Python)

Method Description:
We design a general KD framework for with Attention-based Depth















Distillation for Monocular 3D Object Detection.
Parameters:
1/1/1/0.25/1/0.5
Latex Bibtex:
@article{wu2022attention,







title={Attention-based Depth Distillation with 3D-Aware Positional



Encoding for Monocular 3D Object Detection},







author={Wu, Zizhang and Wu, Yunzhe and Pu, Jian and Li, Xianzhi



and Wang, Xiaoquan},







journal={AAAI2023},





}

Detailed Results

Object detection and orientation estimation results. Results for object detection are given in terms of average precision (AP) and results for joint object detection and orientation estimation are provided in terms of average orientation similarity (AOS).


Benchmark Easy Moderate Hard
Car (Detection) 94.82 % 89.53 % 81.60 %
Car (Orientation) 94.58 % 88.96 % 80.78 %
Car (3D Detection) 25.61 % 16.81 % 13.79 %
Car (Bird's Eye View) 35.20 % 23.58 % 20.08 %
This table as LaTeX


2D object detection results.
This figure as: png eps txt gnuplot



Orientation estimation results.
This figure as: png eps txt gnuplot



3D object detection results.
This figure as: png eps txt gnuplot



Bird's eye view results.
This figure as: png eps txt gnuplot




eXTReMe Tracker