Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

github.com

Post reply on HN
Print view View original

1–10 of 16 posts

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#4

Post by spike021 »

devit wrote: post #3
what-no-tests wrote: post #2

Without access to a crowdsourced database of faces-to-identities this library is next to useless.

Facebook?

Problem is not all photos (profile photos especially) are close-ups of a person's face. Although they could use facial tagging and maybe come up with a composite? I'm probably wrong though.

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#5

Post by dang »

what-no-tests wrote: post #2

Without access to a crowdsourced database of faces-to-identities this library is next to useless.

Please don't post dismissive comments to HN, especially in response to new work.

You have a good point, but there are better ways to put it—e.g. ask if such data is available, or how the author tests the work (this is a Show HN, the author's here)—than a swipe like "next to useless".

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#6

Post by dogma1138 »

spike021 wrote: post #4
devit wrote: post #3

Earlier quoted context omitted.

Facebook?

Problem is not all photos (profile photos especially) are close-ups of a person's face. Although they could use facial tagging and maybe come up with a composite? I'm probably wrong though.

You don't need all photos, you can run a basic face recognition algorithm like the ones used in cameras to identify good candidates and just filter all other photos.

Profile photos are good enough, and with the amount of selfies people are taking there should be more than enough candidates for a full facial recognition matching.

Also facebook isn't the only social network, LinkedIn profile pictures are usually much better for facial recognition Google+ profile pictures are also usually quite good because they crop your face into that silly circle.

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#7

Post by spike021 »

dogma1138 wrote: post #6
spike021 wrote: post #4

Earlier quoted context omitted.

Problem is not all photos (profile photos especially) are close-ups of a person's face. Although they could use facial tagging and maybe come up with a composite? I'm probably wrong though.

You don't need all photos, you can run a basic face recognition algorithm like the ones used in cameras to identify good candidates and just filter all other photos. Profile photos are good enough, and with the amount of selfies people are taking there should be more than enough candidates for a full facial recognition matching. Also facebook isn't the only social network, LinkedIn profile pictures are usually much b…

Oh I see. thanks for the explanation. LinkedIn definitely makes more sense too since more users will have clearer profile shots.

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#8

Post by M4v3R »

Would it help somehow if I provided the authors with some of my private photos for training the network? I have thousands photos that have faces/names tagged on them in my Photos app (OS X). If that would be of any help, I would happily provide some of them.

Re: Show HN: OpenFace – Face recognition with Google's FaceNet deep neural network

#9

Post by jrbaldwin »

spike021 wrote: post #4
devit wrote: post #3

Earlier quoted context omitted.

Facebook?

Problem is not all photos (profile photos especially) are close-ups of a person's face. Although they could use facial tagging and maybe come up with a composite? I'm probably wrong though.

We usually "region propose" and crop to a certain area (in this case the face area, usually at 256x256) then transform to align eye areas before passing to training. This is to standardize the data beforehand. I'm not sure if this lib does region proposal but you can easily write a pre-processor with openCV face plugins to identify face regions (if any, maybe your training image is a landscape not a face!) for cropping.
Post reply on HN