Hm, so fixed version has a worse match rate? Could you please play with FINGER_REMOVED_THRESHOLD and DIFF_THRESHOLD values in libfprint/drivers/upeksonly.c? Try 200-400 for FINGER_REMOVED_THRESHOLD and 5-10 for DIFF_THRESHOLD
On Wed, Feb 10, 2016 at 12:37 PM, Julian Sikorski <beleg...@gmail.com> wrote: > Hi Vasily, > > it works much better now, thank you! I am not getting as many matches as > I would like, but at least the feedback is given immediately. > > Best regards, > Julian > > W dniu 10.02.2016 o 08:06, Vasily Khoruzhick pisze: >> Hi Julian, >> >> Could you please try feb2016-wip branch of >> https://github.com/anarsoul/libfprint ? >> >> Thanks! >> >> Regards, >> Vasily >> >> On Tue, Feb 9, 2016 at 2:01 PM, Julian Sikorski <beleg...@gmail.com> wrote: >>> Attached >>> >>> W dniu 09.02.2016 o 22:24, Vasily Khoruzhick pisze: >>>> So it doesn't detect finger removal. Please apply this patch, >>>> recompile libfprint and make another log. Thanks! >>>> >>>> On Tue, Feb 9, 2016 at 1:12 PM, Julian Sikorski <beleg...@gmail.com> wrote: >>>>> W dniu 09.02.2016 o 20:14, Vasily Khoruzhick pisze: >>>>>> Hi Julian, >>>>>> >>>>>> On Sun, Feb 7, 2016 at 12:30 AM, Julian Sikorski <beleg...@gmail.com> >>>>>> wrote: >>>>>>> W dniu 02.01.2016 o 01:53, Vasily Khoruzhick pisze: >>>>>>>> On Fri, Jan 1, 2016 at 3:41 PM, Julian Sikorski <beleg...@gmail.com> >>>>>>>> wrote: >>>>>>>>> Hi Vasily, >>>>>>>> >>>>>>>> Hi Julian, >>>>>>> >>>>>>> Hi Vasily, >>>>>>>> >>>>>>>>> I should have checked git first, sorry about that. I updated the >>>>>>>>> Fedora >>>>>>>>> package locally to use git source and the scanner is working, thank >>>>>>>>> you! >>>>>>>> >>>>>>>> Thanks a lot for testing! >>>>>>>> >>>>>>>>> The only issue is that is seems somewhat unresponsive - it appears >>>>>>>>> that >>>>>>>>> it takes some time until a fingerprint is processed, plus there is no >>>>>>>>> messsage that matching failed on the login screen: >>>>>>>>> >>>>>>>>> - I swiped my finger >>>>>>>>> - I waited 15 seconds, nothing happened >>>>>>>>> - I swiped again, logged in immediately >>>>>>>> >>>>>>>> Report these issues to gdm developers. libfprint/fprintd/pam_fprintd >>>>>>>> report failures properly, you can check it by invoking >>>>>>>> "fprintd-verify" from console. >>>>>>> >>>>>>> It took me some time but I finally managed to reproduce the issue with >>>>>>> fprintd-verify alone: >>>>>>> >>>>>>> $ fprintd-verify >>>>>>> Using device /net/reactivated/Fprint/Device/0 >>>>>>> Listing enrolled fingers: >>>>>>> - #0: right-index-finger >>>>>>> >>>>>>> It appears that if I swipe my finger only once, fprint will get stuck >>>>>>> here. On the other hand, if I 'wake' the scanner again - by swiping a >>>>>>> different finger for example - the verification will happen immediately. >>>>>>> >>>>>>> Verify result: verify-match (done) >>>>>>> ReleaseDevice failed: Message did not receive a reply (timeout by >>>>>>> message bus) >>>>>> >>>>>> I'll try to reproduce it with my device. Meanwhile could you please >>>>>> compile libfprint with debug logs enabled, >>>>>> start fprintd manually (just kill running daemon with "sudo pkill -9 >>>>>> fprintd" and start a new one in console with "/usr/lib/fprintd"), >>>>>> try to reproduce the issue and then save log and send it to me. >>>>> >>>>> Hi Vasily, >>>>> >>>>> please find attached. After the first swipe fprintd is spamming the >>>>> console with the lines like >>>>> >>>>> upeksonly:debug [row_complete] diff is 236 >>>>> >>>>> Best regards, >>>>> Julian >>>>> >>>>> >>>>>> >>>>>>> Best regards, >>>>>>> Julian >>>>>>> >>>>>>>> >>>>>>>>> Another minor annoyance is that the default keyring is still asking >>>>>>>>> for >>>>>>>>> password which partially defeats the point of logging in with >>>>>>>>> fingerprint as the password needs to be entered anyway. >>>>>>>>> Aside from these two, great work, thank you again! >>>>>>>>> >>>>>>>>> Best regards, >>>>>>>>> Julian >>>>>>>>> >>>>>>>>> W dniu 01.01.2016 o 21:26, Vasily Khoruzhick pisze: >>>>>>>>>> Hi Julian, >>>>>>>>>> >>>>>>>>>> libfprint from git master supports 147e:1001. See >>>>>>>>>> http://cgit.freedesktop.org/libfprint/libfprint/commit/?id=6fc5293e8330e65ed21c0e43a18b3be061933e74 >>>>>>>>>> >>>>>>>>>> Regards, >>>>>>>>>> Vasily >>>>>>>>>> >>>>>>>>>> On Fri, Jan 1, 2016 at 10:25 AM, Julian Sikorski >>>>>>>>>> <beleg...@gmail.com> wrote: >>>>>>>>>>> Hi list, >>>>>>>>>>> >>>>>>>>>>> was there ever any new development for 147e:1001? libfprint-0.6.0 >>>>>>>>>>> does >>>>>>>>>>> not seem to contain these drivers. >>>>>>>>>>> >>>>>>>>>>> Best regards, >>>>>>>>>>> Julian >>>>>>>>>>> >>>>>>>>>>> W dniu 22.03.2014 o 06:46, Julian Sikorski pisze: >>>>>>>>>>>> Hi Vasily, >>>>>>>>>>>> >>>>>>>>>>>> I don't know if it times out every time, but the results look the >>>>>>>>>>>> same >>>>>>>>>>>> from the user perspective. >>>>>>>>>>>> How do I know fprint_demo is ready for enrollment, and that I >>>>>>>>>>>> should >>>>>>>>>>>> continue with re-scanning the finger? >>>>>>>>>>>> >>>>>>>>>>>> Best regards, >>>>>>>>>>>> Julian >>>>>>>>>>>> >>>>>>>>>>>> W dniu 20.03.2014 09:04, Vasily Khoruzhick pisze: >>>>>>>>>>>>> Hi Julian, >>>>>>>>>>>>> >>>>>>>>>>>>> According to log, it looks like one control transfer (write reg) >>>>>>>>>>>>> of >>>>>>>>>>>>> deinit sequence timed out. Does it happen all the time? >>>>>>>>>>>>> >>>>>>>>>>>>> Please note, that with latest changes all imaging devices do 5 >>>>>>>>>>>>> scans >>>>>>>>>>>>> for enrollment. >>>>>>>>>>>>> >>>>>>>>>>>>> Regards >>>>>>>>>>>>> Vasily >>>>>>>>>>>>> >>>>>>>>>>>>> On Thu, Mar 20, 2014 at 9:13 AM, Julian Sikorski >>>>>>>>>>>>> <beleg...@gmail.com> wrote: >>>>>>>>>>>>>> W dniu 16.03.2014 11:45, Vasily Khoruzhick pisze: >>>>>>>>>>>>>>> Hi to all fingerprint scanner users, >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> It's another call for libfprint testing. >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Good news to all UPEK 147e:1001 users - I've added support for >>>>>>>>>>>>>>> that >>>>>>>>>>>>>>> device into upeksonly driver - so it worth testing. It also may >>>>>>>>>>>>>>> work >>>>>>>>>>>>>>> for 147e:1002 device - just try to add its USB IDs into >>>>>>>>>>>>>>> upeksonly.c >>>>>>>>>>>>>>> and use UPEKSONLY_1001 variant for it. >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Also I've improved image assembling utilities for all AuthenTec >>>>>>>>>>>>>>> scanners, so now match rate is a bit better for AES1660 scanner >>>>>>>>>>>>>>> - so >>>>>>>>>>>>>>> all AES1660 are encouraged to test my changes as well. >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> All changes are pushed into master branch of >>>>>>>>>>>>>>> git://github.com/anarsoul/libfprint.git repo. >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> To obtain the code do: >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> git clone git://github.com/anarsoul/libfprint.git >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Use instructions from [1] to compile and test it. Thanks! >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Regards, >>>>>>>>>>>>>>> Vasily >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> P.S. PLEASE KEEP FPRINT MAILLIST IN CC! >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> [1] >>>>>>>>>>>>>>> http://lists.freedesktop.org/archives/fprint/2012-November/000330.html >>>>>>>>>>>>>>> >>>>>>>>>>>>>> Hello, >>>>>>>>>>>>>> >>>>>>>>>>>>>> 147e:1001 does not seem to be working here. When I open >>>>>>>>>>>>>> fprint_demo and >>>>>>>>>>>>>> pick a finger to enroll, the child window shows nothing for a >>>>>>>>>>>>>> while. >>>>>>>>>>>>>> When I scan my finger regardless, a scan will appear at some >>>>>>>>>>>>>> point, but >>>>>>>>>>>>>> with a message "enrollment failed due to bad scan data, press >>>>>>>>>>>>>> cancel to >>>>>>>>>>>>>> continue". Logs attached. >>>>>>>>>>>>>> >>>>>>>>>>>>>> Best regards, >>>>>>>>>>>>>> Julian >>>>>>>>>>> >>>>>>>>>>> _______________________________________________ >>>>>>>>>>> fprint mailing list >>>>>>>>>>> fprint@lists.freedesktop.org >>>>>>>>>>> http://lists.freedesktop.org/mailman/listinfo/fprint >>>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> fprint mailing list >>>>>>>> fprint@lists.freedesktop.org >>>>>>>> http://lists.freedesktop.org/mailman/listinfo/fprint >>>>>>>> >>>>>>> >>>>> >>> > _______________________________________________ fprint mailing list fprint@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/fprint
