Blobscanner - new release v. 0.1-alpha

robdanet

Hi all,
I just released a new version and I have few changes to announce:
The main and only library's class has now two new constructors plus the old one that will go in a future release. Their signature is :

    Detector(PApplet parent, int thrershold)

and

    Detector(PApplet parent)

Of course if you use the second one you need to pass the threshold value via method call:

    setThreshold(int value)

If you need to scan only a given area of the image, you can define it by calling the following method:

    setRoi(int startx, int starty, int width, int height)

this works like the Processing's rect() method, where you pass starting point's coordinates and rectangle's width and height. Please check the examples and the documentation for guidance. I put few links below where it's possible to download the library before it is available through the IDE.

Here you find the zip file with executable, source code and docs:

Blobscanner's website
or
Google-code

The repository was moved to Github:

Blobscanner's repo

Keep an eye on the project's site where I will post tutorials and annuncements.

antonio molinaro

clankill3r

hey, nice you moved to github! :) I made some changes to your library as well like half a year ago. I will branch them later when i have time, that will be in 2/3 months. Your library was still the best and fastest for me :)

clankill3r

The way you have set up the github is really bad. It would be better if the library template was still included. Now it's a lot of work.

robdanet

Hi clankill3r,

first of all forgive me for not answering sooner, to be honest with you I checked a couple of time a week this thread during the last three months, but I never found any question or message, so I kept coming here always less and less...

About the repository, please can you be more specific on how you think I should set it up and can you explain why as it is now, it's more difficult to contribute, as you said in the "issue" you opened on github. Ha! About that, next time please don't open an issue for things like this, use another channel like an email (do you still have my email, have you?). I just added the library.properties file, which I think it's nice to have in development repository anyway. I don't think the same about .jar files or javadoc files. I think this stuff belongs to the distribution archive, which, being myself the only developer of the project, I think it's more simple adding this type of files to it at the moment of a new release. One can argue about the examples files, but it doesn't takes a lot of work for pushing them (or anything else) into the repository, if someone else will join the project.

One last thing. I don't mean to be rude, but you said you made some changes to my library as well like half a year ago, I don't understand how did you made this changes if you forked Blobscanner only seven days ago, also isn't branching something allowed only to contributors?

clankill3r

No problem. Btw, you can bookmark a discusion by clicking the star at the top: That way you get an email when someone post here.

About the files, those ones: build.properties build.xml library.properties

That way a ant build can be made. They still need a little editing for each person but at least then they are already there.

Half a year ago i just downloaded the source and made changes. It was not with version control or anything like that.

And is your email still:

blobdetector.info [ät} gmail

cause there is no website blobdetector.info

Or was that from the other library?

clankill3r

I just did a benchmark test with the same image. With your library i get around 8 frames (i draw boxes around the blobs).

With this library i get 48 frames on the same image: http://www.v3ga.net/processing/BlobDetection/index-page-home.html

It might be good to look into what is taking so much time.

robdanet

Half a year ago i just downloaded the source and made changes. It was not with version control or anything like that.

That's not entirely true. Indeed you created a new repository on github.com called Blobscanner2, pushed my original code into it with some changes, including in the header section of the Detector.java file where you stated that you didn't know if I was still developing the project and that you had though very nicely to take charge of it. I don't remember the exact words, but I remember that at the end of the header you even left a message for me like "antonio please contact me if you see this". Now the repository is deleted but if you google "blobscanner2", you can still get some results like the one in the image, but the link is death because the as I said the repository was deleted.

Now what's my point ? I think your intentions are good, but you approach is totally wrong, not to say childish and inept. So, because of the respect I have for this forum, and for the people who created it, I'm gently asking you to rethink your approach if you still wish to communicate with me here or anywhere else.

Yes, for anything related to Blobscanner, my email address is still blobdetector.info@gmail.com. If you would like to see a website with that domain name, you offer are welcome.

clankill3r

Yeah i remember now. I had it indeed on github but it wasn't a branch. If you knew about the lines where it said trying to get in touch with you then why did you never contact me?

And what is wrong with my approach, and what makes me childish and inept? If it's my post above this one, maybe it sounds a bit negative like "hey this one is faster". But i didn't mean it like that.

To quote myself from the old forum:

I made a method to blob detect a color range so i can use your library again (it's faster then blobDetection library!).

I always loved your library cause of it's speed. And i was thinking maybe there is something wrong atm what drains it's speed.

robdanet

I never contacted you because, less than two months before I found the code on github, you have emailed me for some problems you were experiencing, without mentioning or asking me if I was still running the project.

Returning to what we are here for, the loss of speed is due to the large amount of false positive detected. That image is a real beast and exposes all the weakness of the type of algorithm that the library uses. Also, images like that, need to be preprocessed with some Gaussian blur plus a good dilatation operator. Luckily, blobs like those are quite hard to be found in real life situations.