Method

Ficha Detector [FichaDet]


Submitted on 22 Jun. 2016 14:11 by
Hui Cao (Ficha Inc.)

Running time:0.2 s
Environment:4 cores @ 2.5 Ghz (C/C++)

Method Description:
Learn high-level features for object detection without
deep learning.
This algorithm is targeted for automotive ADAS
Applications without using GPU.
Parameters:
3 models used: 16x40 24x60 20x40
Latex Bibtex:

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) 70.16 % 58.72 % 53.01 %
This table as LaTeX


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




eXTReMe Tracker