13 comments

[ 3.3 ms ] story [ 36.5 ms ] thread
This looks really cool and something that will be fun to play with.

jetpac looks interesting too.

Any idea why they released this framework?

Did you even read the readme.md? 6 lines from the top

"We're releasing this framework because we're excited by the power of this approach for general image recognition, especially when it can run locally on low-power devices. It gives your iPhone the ability to see, and I can't wait to see what applications that helps you build."

No source. I am glad you have a great looking product.
looks great! is the image processing done in the server-side?
The name fits: You've got to have a deep belief if you trust a 3rd Party binary.
Why all the negativity, everyone? This is an impressive little framework - if you need to have the source code, it's not for you, but as-is, it is already useful for - at the very least - personal image recognition projects.

I love his excitement in his blog post and video about it where he trains a program to recognize his cat: http://petewarden.com/2014/04/08/how-to-add-a-brain-to-your-...

The negativity is because it's not open source, and HN is against anything that isn't open source.
Just curious, Is this the best algorithm to use for image recognition? What's a good library to use on the raspberry pi?
recognition and detection are two different things - for detection deep networks are performing as state of the art. Recent for recognition they are also doing well. However, one main difference is that many recognition tasks traditionally have very few images per object/user (especially when you're testing over hundreds or thousands of subjects). DeepFace by facebook showed that deep nets can be useful in recognition but set a rather unreal (at the moment) president that 100s-1000s of images can be used per subject.
So what's the difference between recognition and detection?
Loved this line of code:

NSString *networkPath = ....

if (networkPath == NULL) { // <- :)

Cheers!