Method

kitti crafter [sensekitti]
https://github.com/byangderek/CRAFT

Submitted on 8 Apr. 2016 15:53 by
Bin Yang (University of Toronto)

Running time:4.5 s
Environment:GPU @ 2.5 Ghz (Python + C/C++)

Method Description:
Pre-trained VGG16 network fine-tuned on KITTI object
training set.
Parameters:
Latex Bibtex:
@inproceedings{binyang16craft,
title={Craft Objects from Images},
author={Yang, Bin and Yan, Junjie and Lei, Zhen and Li,
Stan},
booktitle={Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition},
year={2016}
}

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) 90.76 % 90.00 % 81.83 %
Car (Orientation) 47.06 % 44.56 % 41.50 %
Pedestrian (Detection) 80.12 % 67.28 % 62.25 %
Pedestrian (Orientation) 43.55 % 37.50 % 35.08 %
Cyclist (Detection) 81.76 % 72.50 % 64.00 %
Cyclist (Orientation) 46.65 % 42.12 % 36.66 %
This table as LaTeX


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



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



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



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



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



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




eXTReMe Tracker