Method

Deep Continuous Fusion [la] [UberATG-ContFuse]


Submitted on 26 Jul. 2018 21:19 by
Ming Liang (Uber ATG)

Running time:0.06 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
See the paper.
Parameters:
See the paper.
Latex Bibtex:
@inproceedings{Liang2018ECCV,
title = {Deep Continuous Fusion for Multi-Sensor
3D Object Detection},
author = {Ming Liang and Bin Yang and Shenlong
Wang and Raquel Urtasun},
booktitle = {ECCV},
year = {2018}}

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 (3D Detection) 83.68 % 68.78 % 61.67 %
Car (Bird's Eye View) 94.07 % 85.35 % 75.88 %
This table as LaTeX


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



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




eXTReMe Tracker