Method

ACF [ACF]


Submitted on 9 May. 2015 06:39 by
Chaoyang Zhao (National Lab of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences)

Running time:1 s
Environment:1 core @ 3.5 Ghz (Matlab + C/C++)

Method Description:
ACF method with model size of 128x64 pixels
Parameters:
ModelDsPad=[128 64]
Treedepth=4
nWeaks=4096
Latex Bibtex:
@article{Dollar2014PAMI,
author = {Piotr Doll\'ar and Ron Appel and Serge
Belongie and Pietro Perona},
title = {Fast Feature Pyramids for Object
Detection},
journal = {PAMI},
year = {2014},
}

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
Pedestrian (Detection) 59.81 % 45.67 % 40.88 %
Pedestrian (Orientation) 32.23 % 24.31 % 21.70 %
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




eXTReMe Tracker