loading
 
Training Time
Soner Ören from Turkey  [3 posts]
13 year
Hi,
Is it possible to increase the training time in object recoqnition module?

Thanks.
Anonymous 13 year
What is the intended purpose of doing this? Are you not getting good recognition on a particular object? Can you post an image of that object here?

Increasing the training time would not change anything. The 'time' taken to train objects is NOT heuristic like it is in GP or NN algorithms.

STeven.
Soner Ören from Turkey  [3 posts] 13 year
Thanks for the quick response.
We are trying to track an octagonal object using AVM. According to project we must follow the object from different location like from its right or back etc. So we draw 8 eight different patterns to each face of the object so we can understand where our robot is according to object. To do that we trained 3 face at once so we can track the object at all 360 degrees. But  training 3 face takes time turning the object etc. and doing it fast causes low quality on recognition.
Object's image and the script we are using are posted  

Soner

 

program.robo
EDV  [328 posts] 13 year
The training time of AVM is not a problem in this case. If you not have enough time for training on some object then you should just repeat training one more time with the same object (chooses the same object from "Enter object name" dialog window).
EDV  [328 posts] 13 year
By the way, new version of AVM Navigator with "Navigation by map" mode is available now and robot automatically defines his location in this mode:
http://www.roborealm.com/help/AVM_Navigator.php
Soner Ören from Turkey  [3 posts] 13 year
Thanks for your advices.
I will start to work on new module right away.
EDV  [328 posts] 13 year
See "Route Recorder" for details:
http://www.roborealm.com/forum/index.php?thread_id=4094#

This forum thread has been closed due to inactivity (more than 4 months) or number of replies (more than 50 messages). Please start a New Post and enter a new forum thread with the appropriate title.

 New Post   Forum Index