Method

Maximizing the partial AUC score using the ensemble classifier with spatially pooled features [pAUCEnsT]


Submitted on 27 May. 2014 03:14 by
Sakrapee Paisitkriangkrai (The University of Adelaide)

Running time:60 s
Environment:1 core @ 2.5 Ghz (Matlab + C/C++)

Method Description:
Training the ensemble classifier with the following features: LUV, Magnitude, Orientation, spatially-pooled Covariance and LBP.
Parameters:
pAUC [\alpha,\beta] = [0,2^{-4}] and \nu = 2^{-4}
Latex Bibtex:
@INPROCEEDINGS{Paul2014ARXIV,
author = {S. Paisitkriangkrai and C. Shen and A. van den Hengel},
title = {Pedestrian Detection with Spatially Pooled Features and Structured Ensemble Learning},
booktitle = arXiv,
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) 65.84 % 52.88 % 46.97 %
Cyclist (Detection) 50.51 % 34.90 % 30.35 %
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