2021-01-23 17:23
(2021-01-23 17:23)camgirlfinder Wrote:(2021-01-21 21:16)outbackjosiah Wrote: So no chance for older data (pictures)?
No sadly not.
But I have a good news, I just pushed an update which made the search ~20 seconds faster.
The average search should now only take 30s.
(2021-01-26 16:05)Guest Wrote: After latest update to 1/24 data, results seem to be a lot less consistent. Even images that before were getting top match are not even getting right model to show up.
(2021-01-26 21:10)camgirlfinder Wrote:(2021-01-26 16:05)Guest Wrote: After latest update to 1/24 data, results seem to be a lot less consistent. Even images that before were getting top match are not even getting right model to show up.
Can you send me the job ids? Then I could try to find out what is going on.
(2021-01-26 21:33)camgirlfinder Wrote: Okay I see what you mean, I uploaded the example image for Kati3kat and the new results are total garbage:
Example: https://camgirlfinder.net/jobs/Gx6Vs
New: https://camgirlfinder.net/jobs/HtptQ
Looks like I have to remove all images from January for mfc.
But maybe it is also something else because I would expect that SparkleTits still is the best match for that image. Since for that "account" no new images where added since 1st August 2020.
(2021-01-26 23:39)Guest Wrote: https://camgirlfinder.net/jobs/CUkBw
Result kind bad sad
(2021-01-27 20:10)camgirlfinder Wrote:(2021-01-26 23:39)Guest Wrote: https://camgirlfinder.net/jobs/CUkBwLooks like you found a model that is simply not in the database.
Result kind bad sad
Here is a run with the fix: https://camgirlfinder.net/jobs/LqcQY
If anybody knows the name for her please tell me.
I could search the database with that to be sure she is really not in it.
(2021-01-27 22:00)Guest Wrote: she is famous one asian model, she got dozen alternate id and bunch of videos around the net
https://www.camwhores.tv/search/ellaisbe...ahachinse/
(2021-02-03 15:39)metalmog Wrote: Well done! It's a fantastic job. I was able to find a couple of models I used to follow some time ago![]()
Interface is simple, good and easy to use.
I've worked with face recognition once, and I can only imagine how large is your gallery. You talked about updating the database including new faces from last month. Does it work on an incremental way or do you retrain everything again? On my experience with face rec, updating the gallery frequently was a pain.
Another issue I had was regarding the image quality of query and gallery. Since the gallery was built with screencaps, I assume it's better if the query is also a screencap, instead of a profile picture from the model, correct?
As a suggestion, I assume there is a lot of people inputting new images now, searching their favorite models. How about using them as well? You could have feedback from the users about the accuracy of the predictions. Something like, from the 10 closest models of the query submitted there is a checkbox confirming the correct models. In this way you could have new labeled faces (and bonus stats regarding which models are most looked, which websites, etc).
(2021-02-03 10:41)thag Wrote: Thankyou for this site, this has improved very much over the last few weeks, great programming.
(2021-02-03 15:39)metalmog Wrote: I've worked with face recognition once, and I can only imagine how large is your gallery. You talked about updating the database including new faces from last month. Does it work on an incremental way or do you retrain everything again? On my experience with face rec, updating the gallery frequently was a pain.Yes it works incrementally because I do not use a neural network for the prediction. I ran exactly into this problem retraining would take too long.
(2021-02-03 15:39)metalmog Wrote: Another issue I had was regarding the image quality of query and gallery. Since the gallery was built with screencaps, I assume it's better if the query is also a screencap, instead of a profile picture from the model, correct?Yes the site will work better if one also uses screencaps as search image, because the face encodings will be more similar.
(2021-02-03 15:39)metalmog Wrote: As a suggestion, I assume there is a lot of people inputting new images now, searching their favorite models. How about using them as well? You could have feedback from the users about the accuracy of the predictions. Something like, from the 10 closest models of the query submitted there is a checkbox confirming the correct models. In this way you could have new labeled faces (and bonus stats regarding which models are most looked, which websites, etc).Since I do not use a neural network right now, there is less value in collecting user feedback. It would only be useful to detect if I screwed up something. Like a couple of days ago where only CB result where returned
(2021-02-15 22:46)camgirlfinder Wrote: To anyone that still follows this thread
I managed to speed up the search by a factor of 3.
It now takes less than 10 seconds.
Have fun and please write me if you encounter inconsistent or weird behavior.
(2021-02-16 0:26)dracus Wrote: Thank you for this great tool. I was wondering, regarding old stuff, would it be possible to use the extreme amount of content Camgirl Gallery has?
Example: http://thro.bz/image/3EucE
Lately it is acting up a bit. Like the home page doesn't work, you cannot search, etc. Still, the images are there and camgirlvideos.org and camvideos.me are practically living off the collection you can find here.
(2021-02-17 7:19)spike Wrote: great site. Is it possible that you can fix the mobile version of the site. Images only appear as little boxes. if that is possible, cool. Thanks for such a cool site.