Hey HN, this is Yusuf.
I've built a privacy focused macOS app which makes use of a locally running neural network (YuNet), to notify you if other people are looking at your screen. YuNet runs fully on-device with no data leaving your computer.
The app utilises a 230kb facial detection model, which takes images from your webcam and checks for any faces entering the viewing field of your webcam. If the number of faces exceeds the threshold an alert will be shown.
Built with Python + PyQt, the YuNet code comes from OpenCV. Currently it's a macOS app only, however I will be widening access to windows devices soon.
Link + Source code: https://www.eyesoff.app
I'd love your feedback on the app, I look forward to reading your comments on thoughts and future directions you'd like to see!
5 comments
I hope you understand why the market for a "privacy focused" app might not be happy about the idea of giving that program webcam access. Kudos for showing the source, though.
Thanks for the feedback! And yeah I can see that, this was why the model being ran locally was a necessary for EyesOff. Also, I hope the source allows people to trust I dont send any data outside the app
That’s why the model runs locally.
nice, how about being able to add "approved faces" to the dictionary. Kinda like with google doc spell check I can tell it a mis-spelled word is actually ok. So if my friend Bob is on the list and he's detected that doesn't count.
Hey, thanks for the feedback!
This is on the roadmap, along with gaze detection.
https://github.com/opencv/opencv_zoo/tree/main/models/face_r...
This model lets you upload a reference face and then matches to those in images. I think this allows for the "approved faces" function you mentioned.
I suppose a difficulty may arise when we run a few models at the same time, however there is probably a lot of room for efficiency on the table and thanks to the small model size we are already in a good place.