Comment by pospielov | Hacker News Reader

Comment by pospielov | Hacker News Reader

Hey everyone, we are creating a CompreFace - face recognition solution for developers that could be used without a machine learning experience.

It can be used as any paid SaaS like Amazon Rekognition, but it’s totally free. How to use it:

1. Start the service with one command using docker-compose config

2. Use UI to create face collections

3. Use REST API to add faces to face collections and recognize faces

We also have a roles system, so you can control who has access to face collections. We use the FaceNet library under the hood, but we are planning to add support of the InsightFace library. What we are planning to add in the next release:

* Support of face detection and face verification (only face identification for now)

* GPU support

* Support of age and gender detection

We really need your feedback: * Does it make sense for you to use ready-to-use solutions instead of python libraries?

* What more features do you expect from such a solution?

* It would be great if you try to install it - we tried our best to make it as simple as possible, we want to check it on a wide audience

Any other feedback is welcome!

Official website: https://exadel.com/solutions/compreface/

Github link: https://github.com/exadel-inc/CompreFace

Typically an FR system includes some reference for face/head orientation. "Able to detect faces when the orientation is +/y some roll, pitch, yaw of looking directly at the camera", for example. Additionally, a series/set of ROC Curves demonstrating the system against known datasets would be very useful.

We are going to add landmarks information in one of the next releases. I think It could be calculated from this data. But I understand that it's better to add such functionality to CompreFace, I'll add this to our backlog

About comparison with other systems - as I mentioned we use Facenet under the hood. According to their GitHub, the accuracy on LFW data set is 99.65%: https://github.com/davidsandberg/facenet#pre-trained-models

have you considered to switch to https://github.com/deepinsight/insightface as base model?

edit: never mind, you do https://github.com/exadel-inc/CompreFace/blob/master/embeddi...

Yes, we will publish docker-compose configs that uses insightface with some of the nearest releases. We'll still support Facenet as well, so we'll see what library is more popular.

It sounded like the previous comment wasn't asking for orientation detection as a feature. I interpreted the question to be asking under what orientations CompreFace can be expected to work.

Yes, what are the allowed degrees of rotation away from looking at the camera that a face will be detected? And is there a curve demonstrating the FR accuracy as face rotation away from looking directly at the camera increases?

hmm... good question. I haven't found any information for FaceNet that we use. Do you have similar information for any other face recognition system? I would like to see how they calculate it.

First you populate the gallery with the recovered orientation of each face, so the gallery can be segmented by pose, and then the laborious process of humans evaluating the accuracy via generation of a ROC Curve for the gallery. Which can be segmented by face orientation to show the accuracy as orientation varies. People generally hate calculating these because it requires a human to evaluate the FR matches, and mental fatigue set in very quickly.

I'd love to see this in https://www.home-assistant.io/ !

I looked at other custom components, and it looks like it's not so difficult to add this integration.

But it would be better if someone from home-assistant community create it, as they already know how to run and test the system. We will be happy to help them in this case.

I'll still add it to our backlog, I believe IoT is very promising direction for CompreFace, so I would like to support such things in the future.

Reply on news.ycombinator.com