News for hackers. Stuff that matters.
Re:Face Recognition Using Pytorch by briggers on Tuesday September 03, 2019 @01:57AM (#20864019)
I see from the MTCNN code that this repo (like all others I've seen) is still bouncing tensors between GPU and CPU while passing between the P/R/ONets.

So many ML repos make this mistake in pre/post-processing and end up bottlenecked on CPU.

Anyone know of an MTCNN that's been ported to run more or less fully on GPU? (Or even that does batching instead of an image-by-image approach?)

Re:Face Recognition Using Pytorch by cetra3 on Tuesday September 03, 2019 @02:26AM (#20864119)
I have used this before: https://github.com/blaueck/tf-mtcnn which uses tensorflow for all the xNets.

Example in rust: https://cetra3.github.io/blog/face-detection-with-tensorflow...

Re:Face Recognition Using Pytorch by timesler on Tuesday September 03, 2019 @02:49AM (#20864191)
I'm not aware of any implementation with these features, but they are both on the roadmap for the linked repo. Both should also be achievable. Batch processing, in particular, will be a straight-forward change and should result in quite a speed-up. Although it will require the input images to have the same dimensions.
Re:Face Recognition Using Pytorch by briggers on Tuesday September 03, 2019 @08:41AM (#20865921)
Good stuff.

In my experience inputs to MTCNN tend to be full frames, so the uniform dimension requirement is usually met.

Re:Face Recognition Using Pytorch by cnxhk on Tuesday September 03, 2019 @10:30AM (#20866812)
Re:Face Recognition Using Pytorch by timesler on Tuesday September 03, 2019 @11:23AM (#20867397)
That's a good repo. It uses mxnet right? The aim of the repo in the topic was mainly to provide a clean implementation that could slot easily into an existing pytorch workflow.
Re:Face Recognition Using Pytorch by lorepieri on Wednesday September 04, 2019 @04:54AM (#20875263)
Thanks for sharing. I'm working on face recognition with homomorphic encryption, therefore without compromising the user privacy. The bold goal is the first privacy preserving videocamera. If you find this interesting, I would love to chat about it.
Re:Face Recognition Using Pytorch by timesler on Saturday September 07, 2019 @12:22PM (#20904745)
That sounds like a pretty interesting challenge. Happy to chat more - you can get in touch using the contact details on github.
Re:Face Recognition Using Pytorch by s_Hogg on Tuesday September 03, 2019 @01:28AM (#20863910)
Where in the page is the evidence of the speed claim made in the headline for this submission?

Edit: title has now been changed

Re:Face Recognition Using Pytorch by ldulcic on Tuesday September 03, 2019 @07:35AM (#20865517)
Great repo, been using it for a while now! Thank You Tim!
Re:Face Recognition Using Pytorch by timesler on Tuesday September 03, 2019 @11:25AM (#20867425)
Awesome, and thanks for your feedback in the early days. I've made the tracking interface much nicer as a result.