Method

SJTU-HW [SJTU-HW]


Submitted on 6 Sep. 2018 04:44 by
Liangji Fang (Shanghai JiaoTong University)

Running time:0.85s
Environment:GPU @ 1.5 Ghz (Python + C/C++)

Method Description:
None
Parameters:
None
Latex Bibtex:
@article{zsq2018icip,
title={LED: LOCALIZATION-QUALITY ESTIMATION
EMBEDDED DETECTOR},
author={Zhang,Shiquan and Zhao, Xu and Fang,
Liangji and Fei Haiping and Song Haitao},
booktitle={IEEE International Conference on
Image Processing},
year={2018}
}
@article{fang2018small,
title={Small-objectness sensitive detection
based on shifted single shot detector},
author={Fang, Liangji and Zhao, Xu and Zhang,
Shiquan},
journal={Multimedia Tools and Applications},
pages={1--19},
year={2018},
publisher={Springer}
}

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) 90.81 % 90.08 % 79.98 %
Pedestrian (Detection) 85.42 % 74.24 % 69.34 %
This table as LaTeX


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



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




eXTReMe Tracker