object detection, object recognition, learn new models

asked 2014-07-15 02:52:03 -0500

Since there are a lot of questions especially in the field of "object detection" and "object recognition" i'm going to share a recent paper/project that I have recently come across.

ARTOS is the Adaptive Real-Time Object Detection System, created at the University of Jena (Germany). It can be used to quickly learn models for visual object detection without having to collect a set of samples manually. it uses ImageNet, a large image database with more than 20,000 categories. It provides an average of 300-500 images with bounding box annotations for more than 3,000 of those categories and, thus, is suitable for object detection.

A library written in C++ provides the main functionality of ARTOS with a C-style procedural interface, so that it can be easily integrated with any other project.

I haven't tested it yet, but it looks grate!

edit retag flag offensive close merge delete

Comments

I have been reading through the description of this system ARTOS at the link given - this looks fascinating, but on a practical note it is way overkill for what I am trying to ( recognize a simple 2-dimensional target in a camera input stream and track it across video frames ).

The face recognition examples are also way more complex than what I need.

I see there is no tutorial for the Object Recognition yet, either - the ARTOS stuff is real-time interactive adaptive - I need more simple static compiled-offline patterns that can be found in the input video stream. I get to control (somewhat) what the target I'm trying to track looks like, so I'm going for the absolute simplest examples I can get. I will put this together as a tutorial / how-to once I figure it out...

sum kitteh gravatar imagesum kitteh ( 2014-07-16 03:01:27 -0500 )edit

@albertofernandez did you try it?

sturkmen gravatar imagesturkmen ( 2016-03-20 04:38:33 -0500 )edit

Sorry @sturkmen, I haven't tested it yet.

albertofernandez gravatar imagealbertofernandez ( 2016-04-01 09:34:16 -0500 )edit