eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

better approach, imho: * add a variable to the Eigen recognizer, that holds the comparison flag * make it accessible via the get() set() params interface * additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

better approach, imho: * imho:

1. add a variable to the Eigen recognizer, that holds the comparison flag * flag (NORM_L2 vs MORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface * interface
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T) try to abuse the distance param for this, it's needed elsewhere)

better approach, imho:

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs MORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T) DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho:

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs MORM_MALAHANOBIS)NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interfaceinterface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho:imho (since it does not need to modify th FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho (since it does not need to modify th the FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too, and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho (since it does not need to modify the FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer, too,

and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho (since it does not need to modify the FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

again, PLEASE TRY WITHOUT MODYFYING THE INTERFACE!

if your idea (changing the distance func ) turns out nicely, ofc. you want EVERYONE to use it. make it an option as easy to use as possible.

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer,

and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho (since it does not need to modify the FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

again, PLEASE TRY WITHOUT MODYFYING THE INTERFACE!

if your idea (changing the distance func ) turns out nicely, ofc. you want EVERYONE to use it. make it an option as easy to use as possible.

"i got a hammer, now everything looks lika like a nail"

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer,

and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, imho (since it does not need to modify the FaceRecognizer interface):

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

again, PLEASE TRY WITHOUT MODYFYING THE INTERFACE!

if your idea (changing the distance func ) turns out nicely, ofc. you want EVERYONE to use it. make it an option as easy to use as possible.

"i got a hammer, now everything looks like a nail"

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer,

and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach, approach , imho (since it does not need to modify the FaceRecognizer interface):interface),

extend the Algorithm interface:

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

again, PLEASE TRY WITHOUT MODYFYING THE INTERFACE!

if your idea (changing the distance func ) turns out nicely, ofc. you want EVERYONE to use it. make it an option as easy to use as possible.

"i got a hammer, now everything looks like a nail"

eigen, fisher, lbp, they all inherit a virtual method predict(one of 2 actually) method from FaceRecognizer, so, if you insist on your current approach(modifying the interface), you 'll have to modify FaceRecognizer,

and EVERYTHING, that inherits FaceRecognizer.

not recommended.

( oh, and please DON'T try to abuse the distance param for this, it's needed elsewhere)

better approach , imho (since it does not need to modify the FaceRecognizer interface),

extend the Algorithm interface:

1. add a variable to the Eigen recognizer, that holds the comparison flag (NORM_L2 vs NORM_MALAHANOBIS)
2. make it accessible via the get() set() params interface ( so you can switch the behaviour by set() ing a value at runtime)
3. additionally you'll have to modify the serialization ( save() load() )

again, PLEASE TRY WITHOUT MODYFYING THE INTERFACE!

if your idea (changing the distance func ) turns out nicely, ofc. you want EVERYONE to use it. make it an option as easy to use as possible.