Berani

I'm using cv.jit.blobs.bounds in combination with jit.iter to output left, top, right and bottom coordinates of each blob in my window. What I want is to find blobs of a particular size. Next I want to set conditions regarding to this size and if they are met I want to trigger an action.

When I use print this works like a charm, it shows the info of each blob in my max window. But, I can't touch this data and I need it. When I try to use it now, I can only seem to grab the data of 1 blob in my picture which is the data printed in the last line of my max window.

Is there a way to let max constantly display the info of the blobs in my patch itself? I want it to refresh and show info from blob 1 to x (maybe cycle through all the blobs, no matter how fast and just put the data for instance in 4 number boxes with left, top, right and down locations) so I can detect if there are any blobs with dimensions I'm looking for.

What I do now is connect the iter object to a tbl box and print which prints all the position info of each blob. To try and use this data I use a cord from the same iter object, unpack it and let it go to four number boxes. The problem as mentioned earlier is that it will only show the info of 1 blob (last printed line).

detecting-blob-sizes-no-matter-how-many

> What I do now is connect the iter object to a tbl box and print which prints all the position info of each blob. To try and use this data I use a cord from the same iter object, unpack it and let it go to four number boxes. The problem as mentioned earlier is that it will only show the info of 1 blob (last printed line)

the problem is that once you pass the list through the jit.iter it won't 

be a list anymore! It'll be *iterated* out instead, hence the name.

Bypass the iter for everything but the print object, and then unpack it, 

Bart skrev:
> What I do now is connect the iter object to a tbl box and print which prints all the position info of each blob. To try and use this data I use a cord from the same iter object, unpack it and let it go to four number boxes. The problem as mentioned earlier is that it will only show the info of 1 blob (last printed line)
Hi Bart,
the problem is that once you pass the list through the jit.iter it won't 
be a list anymore! It'll be *iterated* out instead, hence the name.

Bypass the iter for everything but the print object, and then unpack it, 
that should work.

> Bypass the iter for everything but the print object, and then unpack it,

Thanks, so how do I get the blobdata (left, rop, right and bottom) directly from cv.jit.blobs.bounds then without using iter?

In an earlier thread I was adviced to use iter to get to this data.

> Bypass the iter for everything but the print object, and then unpack it, 

In an earlier thread I was adviced to use iter to get to this data.


Wow, I've tried a lot of different things, I'm getting dizzy..lol.

What I don't understand is how to get the data out of cv.jit.blobs.bounds without jit.iter.

Each blob has a left top right and bottom. Now this is exactly what jit.iter outputs with print but I'm not sure how to get this info out of cv.jit.blobs.bounds otherwise plus keeping the list and not just the last line.

take a closer look at cv.jit.blobs.bounds.draw, it part of the cvbounds help

file, inside there is a subpatch which passes the cvBounds via [ jit.iter ]

to [framerect $1 $2 $3 $4] message box and from there to jit.lcd.

> Wow, I've tried a lot of different things, I'm getting dizzy..lol.

> What I don't understand is how to get the data out of cv.jit.blobs.boundswithout

> Each blob has a left top right and bottom. Now this is exactly what

> jit.iter outputs with print but I'm not sure how to get this info out of

> cv.jit.blobs.bounds otherwise plus keeping the list and not just the last

take a closer look at cv.jit.blobs.bounds.draw, it part of the cvbounds help
file, inside there is a subpatch which passes the cvBounds via [ jit.iter ]
to [framerect $1 $2 $3 $4] message box and from there to jit.lcd.
also


On 1/23/07, Bart  wrote:
>
>
> Wow, I've tried a lot of different things, I'm getting dizzy..lol.
>
> What I don't understand is how to get the data out of cv.jit.blobs.boundswithout
> jit.iter.
>
> Each blob has a left top right and bottom. Now this is exactly what
> jit.iter outputs with print but I'm not sure how to get this info out of
> cv.jit.blobs.bounds otherwise plus keeping the list and not just the last
> line.
>

The example you posted in your reply is the exact same stuff I use in my patch. It enables me to only grab the last data it gets from cv.jit.blobs.bounds (so, just the data of one single blob)

I'll go check out the help file you mentioned to see if that can help me.

Ok, maybe..(read probably) I'm missing something, but it looks like the examples given are what I'm using already. Let's say that I have four blobs in a window. I want four number boxes that show left top, right and bottom coordinates of blob number 1 in numberbox one to four, next i need the same info in the same boxes but for blob number two and so on.

The print window does exactly the same, it starts with the first blob shows the coordinates on one line and then jumps to the next blob etc. Why don't the numberboxes that are connected to the exact same outlet as the print object, show the updated data of each blob but just the last one?? Is it maybe a timing problem?

The examples provided in previous posts enable me to grab the 4 coordinates, but the problem is that it only displays the data from one blob, even if there are two in a window.

What I need is to cycle through the coordinates of every blob visible just like it does in the print window.

strip all unnecessary stuff from your patch, connect it to the balls example

i think you have a bit of a missunderstanding of max flow of data but ill

> Ok, maybe..(read probably) I'm missing something, but it looks like the

> examples given are what I'm using already. Let's say that I have four blobs

> in a window. I want four number boxes that show left top, right and bottom

> coordinates of blob number 1 in numberbox one to four, next i need the same

> info in the same boxes but for blob number two and so on.

> The print window does exactly the same, it starts with the first blob

> shows the coordinates on one line and then jumps to the next blob etc. Why

> don't the numberboxes that are connected to the exact same outlet as the

> print object, show the updated data of each blob but just the last one?? Is

> The examples provided in previous posts enable me to grab the 4

> coordinates, but the problem is that it only displays the data from one

> blob, even if there are two in a window.

> What I need is to cycle through the coordinates of every blob visible just

strip all unnecessary stuff from your patch, connect it to the balls example
from cv.jit.sort and post.
i think you have a bit of a missunderstanding of max flow of data but ill
need a patch to help you further.

On 1/24/07, Bart  wrote:
>
>
> Ok, maybe..(read probably) I'm missing something, but it looks like the
> examples given are what I'm using already. Let's say that I have four blobs
> in a window. I want four number boxes that show left top, right and bottom
> coordinates of blob number 1 in numberbox one to four, next i need the same
> info in the same boxes but for blob number two and so on.
>
> The print window does exactly the same, it starts with the first blob
> shows the coordinates on one line and then jumps to the next blob etc. Why
> don't the numberboxes that are connected to the exact same outlet as the
> print object, show the updated data of each blob but just the last one?? Is
> it maybe a timing problem?
>
> The examples provided in previous posts enable me to grab the 4
> coordinates, but the problem is that it only displays the data from one
> blob, even if there are two in a window.
>
> What I need is to cycle through the coordinates of every blob visible just
> like it does in the print window.
>
> Is this making any sense?
>

> What I need is to cycle through the coordinates of every blob visible just like it does in the print window. 

But doesn't the max window print all of the blobs in the blink of an 

eye, because of the fast iteration? That would mean that the data *is* 

getting into the numberboxes, but iter just gets to the last matrix of 

data too quickly, thus only showing the last one long enough to see.

(at least, that was the point I was trying to make in my previous post, 

Bart skrev:
> What I need is to cycle through the coordinates of every blob visible just like it does in the print window. 
>
But doesn't the max window print all of the blobs in the blink of an 
eye, because of the fast iteration? That would mean that the data *is* 
getting into the numberboxes, but iter just gets to the last matrix of 
data too quickly, thus only showing the last one long enough to see.

(at least, that was the point I was trying to make in my previous post, 
sorry if I wasn't clear enough on that.)
Cheers,
Andreas


You were clear, I'm just still too much of a noob to get it, but it's starting to make sense now with the extra info you gave me.

Actually I experienced that all of the data gets through the boxes, although it does it too fast. When I tried to do some calculations, it actually used data from different blobs while this wasn't the intention. I guess what you see is not always realtime.

Anyway, I'll try to post something to make it easier to understand.

Ok, here's what I'm using connected to the sort patch as requested.

Now, as you can see I get data of all the blobs in the print window.

Also, I can get the right data per blob but this would require me knowing how many there are, which I don't (in the example you see 2). The unpack object in the lower right part of the screen actually gets the same data as the print window shows..but as mentioned earlier here's where my problem comes up, it shows just one line. I need that part to show the data of each blob one by one. If that's possible of course..

okay Bart, I had a quick look and I found a quick hack solution for you.

Basically, instead of using iter, which doesn't do anything in terms of 

preserving the order of the blobs, try connecting these to you 

Basically, instead of using iter, which doesn't do anything in terms of 
preserving the order of the blobs, try connecting these to you 
jit.cv.blobs.bounds left outlet:

> Basically, instead of using iter, which doesn't do anything in terms 

> of preserving the order of the blobs...

ahem. I didn't just say that - jit.iter outputs the plane through the 

middle outlet - that works like an index, really.

Andreas Wetterberg skrev:
> Basically, instead of using iter, which doesn't do anything in terms 
> of preserving the order of the blobs...
ahem. I didn't just say that - jit.iter outputs the plane through the 
middle outlet - that works like an index, really.
A.


> okay Bart, I had a quick look and I found a quick hack solution for you.

this is a LOT more elegant, I think - note that the middle outlet of the 

jit.iter is used to select where the lists go:

Andreas Wetterberg skrev:
> okay Bart, I had a quick look and I found a quick hack solution for you.
and boy, was that a hack...

this is a LOT more elegant, I think - note that the middle outlet of the 
jit.iter is used to select where the lists go:

But, this is still showing data per blob, which is cool but what if I have 199 blobs?

Isn't it possible to let each line of info (left, top, right, bottom) per blob go through 1 set of four number boxes? This way i can look for certain values in each blob.

cv.jit.blobs.bounds outputs the info of all visible blobs right?

Can I use this info and read it in 4 number boxes just like the print info but cycling through each blob from number 1 to number x?

Right now the patch you gave me actually shows the same info as I had for the first two blobs. But I don't want to seperate it per blob. Dang, my English prevents me from explaining what I mean..lol.

O wait, I see you posted more, let me check that first..sorry.

Ok, in your example you split up the values of each blob to three different sets of number boxes.

But, is it also possible to just let all of the blob values go through 1 set of number boxes? So it starts with values of blob 1 till blob x.

Your solution is way more efficient indeed btw..:)

> Ok, in your example you split up the values of each blob to three different sets of number boxes.

> But, is it also possible to just let all of the blob values go through 1 set of number boxes? So it starts with values of blob 1 till blob x.

Adjust the timing on the metro, stop metro and tweak numbox to select 

Bart skrev:
> Ok, in your example you split up the values of each blob to three different sets of number boxes.
>
> But, is it also possible to just let all of the blob values go through 1 set of number boxes? So it starts with values of blob 1 till blob x.
>
Here you go:
Adjust the timing on the metro, stop metro and tweak numbox to select 
manually.

Wow, that's impressive..lol. Thanks for your input guys, I'm gonna have to mess around with these for a while because now only the first row of jit.cellblock shows data of a blob..wonder why..

But I definitely think it can solve my problem once I figure out how it works.

Btw, is it true that the jit.cellblock example uses a subpatch?

Quote: Berani wrote on Thu, 25 January 2007 12:59

----------------------------------------------------

> Wow, that's impressive..lol. Thanks for your input guys, I'm gonna have to mess around with these for a while because now only the first row of jit.cellblock shows data of a blob..wonder why..

That's because there's only one blob in your image. It works fine for me. Have you forgotten to use cv.jit.label before cv.jit.blobs.bounds?

And, yes, there is a sub-patch. Double click on the object [p pad_data] to see what's inside. As your patches get bigger, using sub-patchers becomes pretty much a necessity. Here, I could have done without one, be it helps label the program better.

Quote: Berani wrote on Thu, 25 January 2007 12:59
----------------------------------------------------
> Wow, that's impressive..lol. Thanks for your input guys, I'm gonna have to mess around with these for a while because now only the first row of jit.cellblock shows data of a blob..wonder why..
>

And, yes, there is a sub-patch. Double click on the object [p pad_data] to see what's inside. As your patches get bigger, using sub-patchers becomes pretty much a necessity. Here, I could have done without one, be it helps label the program better.


I connected cv.jit.label @charmode 1 @threshold 400 to cv.jit.blobs.bounds, but it only fills the first row of cells.

The patch of Andreas really comes close to what I need, but because like you said number boxes aren't really meant to handle data that's spitted out constantly, it's causing some unwanted data to slip through every now and then..

I would like to give the jit.cellblock another try though, once I can manage to fill up the cells with data of all blobs visible.

I found the problem, a backslash character didn't show up properly in the patch I pasted. This one should work:

(If it still doesn't work, the message box inside the sub-patcher should be: [dstdimstart 0 0, dstdimend $1 0] )

Now, let's say I would like to search the cellblock for specific values.

Like left should be 4, top 8, right 15 and bottom 20.

The found match of course can be found in different rows.

Now with number boxes I'm a bit more familiar so I found a way to find a match based on some conditions once cycling through all the blobs worked, but I wonder how I can achieve this by using the cellblock patch you created.

I would really like to use the cellblock version because this way it shows all the needed data at once.

What objects would you advice me to use?
The found match of course can be found in different rows.

I would really like to use the cellblock version because this way it shows all the needed data at once.


I suggest you really, really spend more time trying to understand how data flow works in Max. The answer to your question is very simple, but you need a better grasp of how Max works at the base.

As a starting point, I'll just say that the cell block is only for displaying data.

I suggest you really, really spend more time trying to understand how data flow works in Max. The answer to your question is very simple, but you need a better grasp of how Max works at the base. 

As a starting point, I'll just say that the cell block is only for displaying data.


detecting blob sizes, no matter how many