If you’re a programmer, you might have heard of OpenCV before. OpenCV is a machine learning library, that has over 2500 algorithms, and interfaces to C/C++, Python, C#, and so on.
Orange trainer is just a helper for OpenCV, when you want to create samples from your images. I had to work on a project, that needed to recognize an object in real-time, from the webcam. To achieve this, I’ve used OpenCV. At a point, you will need to create an info file, that contains all your positive (with the object) images and also the number of objects in the image and their coordinates. Here is an example:
<em>pos/output3+0105_184_107_157_139.png</em> 1 184 107 157 139
The first part is the image filename. The first number (1), is the number of the objects in the picture. The orange trainer will work only with the images that have one objects in them, not more. 187 and 107 is the x and y, of the left upper corner of the object (in image). 157 and 139 are the width and height of the object (again, in the image).