> Reverse geocoding and automated tagging based on Google TensorFlow.
> ... No privacy concerns.
Tried too look up this. I thought TensorFlow was some hosted Google service but as I understand it now it is an open source library which makes more sense given the quotes above.
Can someone who's more into TensorFlow than me explain this?
It’s, first, an open source library. It optionally integrates with Google Cloud services to use their hardware. But that’s not even likely to be useful in this case, because labeling a few thousand photos just doesn’t require that much power (as opposed to training).
Thanks, so what your saying is this can be done locally, and probably shoukd since there's no reason to involve Google cloud for small data like a local photo collection?
The development of every commercial product is focused on monetization. We've built similar apps more than once and every single time the constraints of working in a profit-oriented corporate environment were an impediment. We are sure we can do better with only a fraction of the budget. Simplicity can be very powerful.
Cool - but is it static? photog is the only photo gallery generation software I have used that will handle large photo galleries efficiently (i.e. it'll work even on a Pi), and that's because it cheats by pre-generating everything.
I understand that "docker run blah" might be comfortable or even preferable for a lot of people. But, for an open source project written in Go(capable of spitting out single executable binary), I would have loved to see plain old build instructions without Docker. as in Download this, and connect to a MySQL database like this.
Not everyone is comfortable with running Docker for several reasons.
If you look at the Dockerfile you will realize this project uses a lot of stuff beyond just the go binary; it looks like docker is a good fit for this project.
Saw this being mentioned at the Go conference in Stockholm last week. Definitely following the progress of this project, seems very useful in times where ad supported image hosting sites close down from time to time.
I always thought a good Google photos alternative was necessary. I have a collection ok 100k+ photos in it, it's great for storage and some sorts of searches, but very limited in some other aspects. And it's impossible to get a bug fixed (like the fact that the mobile app clogs my phone with gigabytes of thumbnails cache) or a feature implemented (grouping albums together when similar faces appear together repetitively). I'm just a front end guy, but i guess it's time to learn go ^^
17 comments
[ 2.9 ms ] story [ 40.2 ms ] thread> ... No privacy concerns.
Tried too look up this. I thought TensorFlow was some hosted Google service but as I understand it now it is an open source library which makes more sense given the quotes above.
Can someone who's more into TensorFlow than me explain this?
Why this has to be free software
The development of every commercial product is focused on monetization. We've built similar apps more than once and every single time the constraints of working in a profit-oriented corporate environment were an impediment. We are sure we can do better with only a fraction of the budget. Simplicity can be very powerful.
Not everyone is comfortable with running Docker for several reasons.