Originally published atwww.indiehackers.com
Offline first iOS camera app with a ML model that detects nudity at about 30-40fps
We have an iOS prototype with our ML model that detects sexually explicit content in realtime. For the geeks out there, we are predicting multiple objects (example- exposed female genitalia, exposed buttocks) at an average of 30 frames a second.
Here is a screen grab from the iOS app prototype 📱: https://ibb.co/k8Z3Tzz
The app can work offline with the help of 2 machine learning models,
- Object detector for realtime recognition.
- Classifier model, to run a final check after an image is captured and before it's saved on to the device.
This gives us extra surety that clicking sexually explicit images would be next to impossible. 😸
The prototyped model is based on an open-source image dataset. We look forward to introduce a lot improvements that further refines our model.
For more information & to signup for our closed beta, visit https://www.getnuno.com/?lang=en
We hope to work on this app and solve a potentially dangerous problem due to the over exposure of online consumption that children/teens these days are introduced into.
If you are interested, and preferably a parent who wishes to help make this app better for the upcoming generation of children, we would be glad to have you join our closed beta program! 🤩