Show HN: Open source Deep Learning Background removal service

[burgalon]

Me and my colleague have been working on a small POC for Background removal. We used semantic segmentation trained on selfies, and documented the process of training and deploying a Keras model with Docker, Heroku and a Gitlab CI in 3 posts:

https://greenscreen-ai.boorgle.com Server side implementation https://greenscreen-ai-client.boorgle.com Client side implementation using KerasJS

Training post https://medium.com/@gidishperber/background-removal-with-dee...

Deploying Keras model to Heroku with Gitlab CI https://medium.com/@burgalon/deploying-your-keras-model-3564...

KerasJS client side post https://medium.com/@burgalon/deploying-your-keras-model-usin...

Server side source code https://gitlab.com/fast-science/background-removal-server

Client side post https://gitlab.com/fast-science/background-removal-vue

[ldenoue]

When I tried the client side version on the same image, it wasn't good compared to the server side. Are you using a different model on the client?

[burgalon]

Indeed I had a problem converting the latest model to KerasJS

[isharko]

This is really cool! Thanks for sharing.

[inmean]

very nice concept, too bad the greenscreen-ai-client a bit slow

[burgalon]

Yep. I'm looking forward to finish our implementation using WebDNN. It's super fast