openbr models don't exist! about openbr

[jklontz]

@bklare @mburge I'd really like you to reconsider releasing the current trained model files. People are going to run into trouble building and using the system otherwise. More importantly though, I firmly believe fostering a community by providing a system that actually works out of the box is of paramount importance right now given the infancy of the project and the need for new contributors. Otherwise I'm made to look like a liar when I give presentations on the system and advertise working algorithms. Certainly it will be important to provide tutorials on training the system on open source datasets, but for the time being we need to provide a system that just works.

[bklare-zz]

While I like Mark's idea of providing the improved models only for
higher tier (i.e., paid) support, the points you just raised are
important. Opportunities to financially leverage this system will
still exist down the road through other improvements avenues. Further,
there is a risk of OpenBR not being adopted if the accuracy is too low
(as jklontz mentioned).

Though mburge has the final say, I do support the release of the
model. It is important to promote growth, we are not yet in a position
to pursue such cost-based support/important, and this is (hopefully)
small potatoes compared to some of the other improvements I plan to
build on this framework.

On 1/10/13, jklontz [email protected] wrote:

@bklare @mburge I'd really like you to reconsider releasing the current
trained model files. People are going to run into trouble building and using
the system otherwise. More importantly though, I firmly believe fostering a
community by providing a system that actually works out of the box is of
paramount importance right now given the infancy of the project and the need
for new contributors. Otherwise I'm made to look like a liar when I give
presentations on the system and advertise working algorithms. Certainly it
will be important to provide tutorials on training the system on open source
datasets, but for the time being we need to provide a system that just
works.

---

Reply to this email directly or view it on GitHub:
#2 (comment)

Brendan F. Klare
Pattern Recognition and Image Processing Lab
Dept of Computer Science and Engineering | Michigan State University
Email: [email protected] | Phone: [phone] |
http://www.cse.msu.edu/biometrics/

[mburge]

Lets release it. Mark

On Thursday, January 10, 2013, bklare wrote:

While I like Mark's idea of providing the improved models only for
higher tier (i.e., paid) support, the points you just raised are
important. Opportunities to financially leverage this system will
still exist down the road through other improvements avenues. Further,
there is a risk of OpenBR not being adopted if the accuracy is too low
(as jklontz mentioned).

Though mburge has the final say, I do support the release of the
model. It is important to promote growth, we are not yet in a position
to pursue such cost-based support/important, and this is (hopefully)
small potatoes compared to some of the other improvements I plan to
build on this framework.

On 1/10/13, jklontz <[email protected] <javascript:_e({}, 'cvml',
'[email protected]');>> wrote:

@bklare @mburge I'd really like you to reconsider releasing the current
trained model files. People are going to run into trouble building and
using
the system otherwise. More importantly though, I firmly believe
fostering a
community by providing a system that actually works out of the box is of
paramount importance right now given the infancy of the project and the
need
for new contributors. Otherwise I'm made to look like a liar when I give
presentations on the system and advertise working algorithms. Certainly
it
will be important to provide tutorials on training the system on open
source
datasets, but for the time being we need to provide a system that just
works.

---

Reply to this email directly or view it on GitHub:
#2 (comment)

Brendan F. Klare
Pattern Recognition and Image Processing Lab
Dept of Computer Science and Engineering | Michigan State University
Email: [email protected] <javascript:_e({}, 'cvml', '[email protected]');>| Phone: [phone] |
http://www.cse.msu.edu/biometrics/

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/2#issuecomment-12103900.

[jklontz]

Done. Thanks for the consensus.

@bhh1988 it should work now. Sorry for the trouble.