##Notes The paths are all relative so it matters where the files are located. Don't need to worry about including CUDACopyTo or CUDACopyFrom in the pipeline; they are automatically included.
##Files
###bashEnroll.sh
- use for enrolling
- right now, there is no gallery output
- example: ./bashEnroll.sh -a CPUCVT+CUDALBP+CUDAPCA -n 1,2,3 -e /data/MEDS/img/
- example: ./bashEnroll.sh -a trainedAlgFile.alg -n 1,2,4
Flags | Required | Parameter | Desc |
---|---|---|---|
a | yes | name of the trained algorithm or pipeline of plugins | specify the trained algorithm file or pipeline to enroll; the pipeline can use key words |
b | no | relative path to 'br' executable | default path is "openbr/build/app/br/br" |
e | no | directory of images to enroll with | default database is "/data/ATT/img/" |
g | no | name of the outputted gallery file | default base name is enrolledGallery.gal; the image dimensions, type of algorithm, and number of copies used for enrolling get incorporated into the gallery name; final name follows convention: [name inputted][database image dimensions][CUDA/CPU type][number of copies for training][number of copies for enrolling].[extension] |
h | no | help | |
n | yes | comma separated sequence of numbers | specify the number of copies of the enrolling image directory to use; numbers need to be in increasing order |
o | no | output filename for timing data | default name is "timingEnroll.csv" |
###bashTrain.sh
- use for training
- example: ./bashTrain.sh -p CPUCVT+CUDALBP+CUDAPCA -n 1,2,3 -t /data/MEDS/img/
Flags | Required | Parameter | Desc |
---|---|---|---|
a | no | name of the trained algorithm | default base name is trainedAlgorithm.alg; final name follows convention: [name inputted][database image dimensions][CUDA/CPU type]_[number of copies].[extension] |
b | no | relative path to 'br' executable | default path is "openbr/build/app/br/br" |
h | no | help | |
n | yes | comma separated sequence of numbers | specify the number of copies of the enrolling image directory to use; numbers need to be in increasing order |
o | no | output filename for timing data | default name is "timingTrain.csv" |
p | yes | specify the algorithm pipeline either with keywords or have full pipeline inside of quotes | |
t | no | directory of images to train with | default database is "/data/ATT/img/" |
###Key Words
Word | Meaning |
---|---|
CPUCVT | Use the CPU version of the Cvt plugin: "Cvt(Gray)" |
CPULBP | Use the CPU version of the LBP plugin: "LBP" |
CPUPCA | Use the CPU version of the PCA plugin: "PCA" |
CUDACVT | Use the CUDA version of the Cvt plugin: "CUDARGB2GrayScale" |
CUDALBP | Use the CUDA version of the LBP plugin: "CUDALBP" |
CUDAPCA | Use the CUDA version of the PCA plugin: "CUDAPCA" |