Forum

Object Detection Deeper image understanding

vvvv beta question
#1

Hello Forum,

currentlty I want to start experiments using object recognition in images. Simple example: This is a bycicle, this is a lamp…I found this article from google "Building a deeper understanding of images ".
http://googleresearch.blogspot.de/2014/09/building-deeper-understanding-of-images.html
My question is: Are there nodes in VVVV to experiment with. I know the Detect Object node, but maybe this is not flexible enough or there are further developments here

https://vvvv.org/sites/default/files/imagecache/large/images/Unbenanntes%20Bild.png

I am very pleased about your feedback

Greetings from Stuttgart
Roman

#2

Maybe @motzi’s vvvv.packs.machinelearning contrib?

Links to contributions not translated correctly
#3

Hey Bjoern,

thank’s lot for your fast response. Looks very interresting and also very close to my wishes. There are a lot of nodes available. I’m not that deep into this topic. Do you have an idea where to start from?

Greetings from Stuttgart
Roman

#4

for a decent programmer it should be not too hard to implement a vvvv plugin for this library: https://vimeo.com/146492001

in the video description you can find the links to the sources. maybe the training files can be used by motzis machine learning framwork.

#5

Sorry, other than a very brief introduction by @motzi I’ve got no experience with the pack whatsoever. The Accord.NET website might be a good starting point though (e.g. by trying to correlate the examples with the existing nodes)

#6

@tonfilm & @ bjoern: Thanks a lot for your help. The video is nice :)…never seen before that fast. I will try and give respronse if I get something to work.

© vvvv group