This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
by default, the system is setup for identification (retrieve the classid of closest train image).
if you want to use it for age or gender recognition, you'll have to relabel your train data, i.e. [0 1] labels for gender.
if you're using csv files for this, you'll have to change it, assuming 0 is male and 1 is female:
at/s13/2.pgm;0
at/s13/7.pgm;0
at/s13/6.pgm;0
[...]
at/s17/5.pgm;0
at/s17/3.pgm;0
[...]
at/s8/1.pgm;1
at/s8/2.pgm;1
at/s8/3.pgm;1
[...]
at/s10/1.pgm;1
ofc. you can use your own images, just add a line with the image path, a semicolon, and the gender id.
| | 2 | No.2 Revision |
by default, the system is setup for identification (retrieve the classid of closest train image).
if you want to use it for age or gender recognition, you'll have to relabel your train data, i.e. [0 1] labels for gender.
if you're using csv files for this, you'll have to change it, assuming 0 is male and 1 is female:
at/s13/2.pgm;0
at/s13/7.pgm;0
at/s13/6.pgm;0
[...]
at/s17/5.pgm;0
at/s17/3.pgm;0
[...]
at/s8/1.pgm;1
at/s8/2.pgm;1
at/s8/3.pgm;1
[...]
at/s10/1.pgm;1
ofc. you can use your own images, just add a line with the image path, a semicolon, and the gender id.
then, you'll have to re-train it using this data, and it will predict 0 or 1 labels later, instead of person-id's.