loading
 
Automating the learn object process in AVM navigator
Kamlesh Thakur from India  [32 posts]
5 years
Hello,

I am using the AVM navigator v0.7.4.1. I want to automate the learn object process for a detected object. The main objective is to train the AVM navigator with human eye object automatically when the eye objected is detected from the human face. Please see the attached file. I have used 'Canny 0-30' and 'circles' commands from RR to detect the circular human eye puppil. As a next step, I want to automatically set the key image size to fit the detected eye object within and start the 'learn object' process automatically.

Request you to provide me the API's or RR program commands available to execute these steps.

Best Regards,
Kamlesh Thakur

 
EDV  [328 posts] 5 years
Navigator package is updated now and you can download next modification of AVM Navigator v0.7.4.2 from your account link.

In this version API for writing the new images to AVM search tree was added.

Use the variables "NV_WR_OBJ_RECT_X, NV_WR_OBJ_RECT_Y, NV_WR_OBJ_RECT_W,
NV_WR_OBJ_RECT_H" for setting of object rectangle coordinates.

NV_WR_OBJ_RECT_X - left-top corner X coordinate of object to be recognized
NV_WR_OBJ_RECT_Y - left-top corner Y coordinate
NV_WR_OBJ_RECT_W - width
NV_WR_OBJ_RECT_H - height

NV_WR_OBJ_IDX - the object index for writing

This variable provides single writing to AVM with object index
that was specified by user and further it will be switched to -1.

The object index just associates with object name that
was given by user during object learning
(button "Learn object" in "Object recognition" mode):

Input image -> AVM search tree -> Object index ->
Table of object names -> Object name for print below object's rectangle

And "Table of object names" will not be cleared even after clicking on
"Set key image size (New)" that clears all object recognition data.

So if AVM search tree already has recognition data that was associated
with this object index then new data just will be added and also associated
with the same object index. And when you will show opposite your camera old
and new objects then both objects will be recognized and you will see
the same name below objects.

Here is an example of using for this feature:
program.robo

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