Testing the water to see if there is interest in a component:
Appears to be one of the fastest cloud services. Will make use of this package.
Making progress, and impressed how fast the API is, 0.5 sec to identify a face + emotions
Sounds cool, might make a good facial recognition addition - emotions could be a cool addition.
2 Likes
Yes, very useful add-on for CCTV cameras! Meta tag humans, animals, burglers… Also sort number plates etc. Very very cool!
2 Likes
That auth with google services is a pain, haven’t done much with them to date
Thats to bad… Think what a neural network API would do for home automation, the house learns when lights should turn on etc. For home use it should be at no cost when you look at the price list…
It’s easy enough to implement, you should have a go 
This will get you started
2 Likes
Hi Robin,
did you ever complete the Google Cloud Vision component?
I’m trying to build a consensus system to detect persons without false positives and so far TensorFlow and your Rekognition components are doing a good job (except for a . However, I’d like to add a third algo and DeepStack is very far behind to be useful in my particular setting (only detects people if very clearly visible). Can you recommend anything else?
Hi @jiiins
re your deepstack comment, perhaps low accuracy is related to your specific use case. It is using Yolov3 which has good accuracy in benchmarking tests.
Re Google vision, I didn’t put much effort into it, and didn’t really like the API. But its definitely worth including in any comparison.
Cheers
I have started work on this integration, basic functionality there, will fill it out over next few days