Jeremy Janella
75 lines
2.4 KiB
Plaintext
75 lines
2.4 KiB
Plaintext
**********************************************
|
|
The German Traffic Sign Recognition Benchmark
|
|
**********************************************
|
|
|
|
This archive contains the test set of the
|
|
"German Traffic Sign Recognition Benchmark".
|
|
|
|
This is the official test set that is used during the final
|
|
competition session at IJCNN 2011. After the conference,
|
|
it will be published on our website.
|
|
|
|
If you want to publish results based on GTSRB,
|
|
this data set is to be used for final testing purposes
|
|
only. Please use the GTSRB (final) training set for training
|
|
(It is the joint dataset of the online competition's
|
|
training and test set).
|
|
|
|
|
|
**********************************************
|
|
Archive content
|
|
**********************************************
|
|
This archive contains the following structure:
|
|
|
|
There is one directory that contains all test images (12,630 images)
|
|
and one text file with annotations (GT-final_test.test.csv)
|
|
|
|
|
|
**********************************************
|
|
Image format and naming
|
|
**********************************************
|
|
The images are PPM images (RGB color). Files are numbered in ascending order:
|
|
|
|
00000.ppm to 12629.ppm
|
|
|
|
Images are in random order, ie. there is no track structure or class information
|
|
as in the training set.
|
|
|
|
**********************************************
|
|
Annotation format
|
|
**********************************************
|
|
|
|
The annotations are stored in CSV format (field separator
|
|
is ";" (semicolon) ). The annotations contain meta information
|
|
about the image and the class id.
|
|
|
|
|
|
In detail, the annotations provide the following fields:
|
|
|
|
Filename - Image file the following information applies to
|
|
Width, Height - Dimensions of the image
|
|
Roi.x1,Roi.y1,
|
|
Roi.x2,Roi.y2 - Location of the sign within the image
|
|
(Images contain a border around the actual sign
|
|
of 10 percent of the sign size, at least 5 pixel)
|
|
|
|
**********************************************
|
|
Further information
|
|
**********************************************
|
|
For more information on the competition procedures and to obtain the test set,
|
|
please visit the competition website at
|
|
|
|
http://benchmark.ini.rub.de
|
|
|
|
If you have any questions, do not hesitate to contact us
|
|
|
|
tsr-benchmark@ini.rub.de
|
|
|
|
|
|
**********************************************
|
|
Institut für Neuroinformatik
|
|
Real-time computer vision research group
|
|
|
|
Ruhr-Universität Bochum
|
|
Germany
|
|
********************************************** |