Method

yolo4 [yolo4]


Submitted on 19 May. 2020 05:27 by
zhiping wang (xjtu)

Running time:0.02 s
Environment:1 core @ 2.5 Ghz (C/C++)

Method Description:
yolov4
Parameters:
width=416
height=416
channels=3
momentum=0.949
decay=0.0005
angle=0
saturation = 1.5
exposure = 1.5
hue=.1

learning_rate=0.001
burn_in=1000
max_batches = 6000
policy=steps
steps=4800,5400
scales=.1,.1
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
Car (Detection) 94.71 % 90.63 % 80.38 %
Pedestrian (Detection) 72.49 % 55.78 % 51.11 %
Cyclist (Detection) 67.33 % 48.67 % 43.00 %
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



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




eXTReMe Tracker