|
Class based object recognition
Trop from Switzerland [6 posts] |
12 year
|
I realize that the object recognition module in roborealm uses each template image as an object to recognize, however how would i use this module to track two objects each with multiple training images.
For example, I would like to detect faces(i have say 200 training samples), and I would like to detect cars (i have another 200 training samples). Currently if i train 200 face images, and 200 car images, roborealm recognized the closest template, and gives the filename, but I would just like to find out if its a face, or a car.
Would I have to have 2 instances of this module running, one for each class of object I would like to recognize?
|
|
|
|
Anonymous |
12 year
|
|
Couldn't you just build that into the name? Like Ford.car, sam.person?
|
|
|
|
Anonymous |
12 year
|
As anonymous points out, you can also have two folders in the training folder (car and face) and then just check the last folder of each found object. The entire path of the found object is recorded exactly for this purpose (i.e. classification/clusters/etc).
STeven.
|
|
|
|
Trop from Switzerland [6 posts] |
12 year
|
|
Thanks for your answers, I will give that a go :)
|
|
This Thread