11.7: Computer Vision: Blob Detection - Processing Tutorial

In this computer vision tutorial, I build on top of the color tracking example and demonstrate a technique known as "blob detection" to track multiple objects of the same color.
Support this channel on Patreon: / codingtrain
Send me your questions and coding challenges!
Contact: / shiffman
Links discussed in this video:
Computer Vision for Artists and Designers by Golan Levin: www.flong.com/t...
Image Processing in Computer Vision: openframeworks....
Source Code for the Video Lessons: github.com/Cod...
p5.js: p5js.org/
Processing: processing.org
For More Computer Vision videos: • Computer Vision
For More Coding Challenges: • Coding Challenges
Help us caption & translate this video!
amara.org/v/QbrM/
📄 Code of Conduct: github.com/Cod...

[@boiimcfacto2364]

I know I say this a lot in the comments, but holy hell was this video incredible! I'm watching it 5 years later, and it just helped me develop a large scale project. These videos are a treasure trove of information Dan, love you for these!

[@finalfantasy7820]

I can't believe you just made me watch a 21-min video about something I already know.

[@furrane]

The blobs attack at the end was epic haha :) Great stuff as always :)

[@TheCodingTrain]

Heh, thanks!

[@dhananjaysonawane1996]

A perfect teacher! There is algorithm + code + corner cases + customization + dance + most important fun !! Liked, subscribed and I am going to explore your channel!! Thanks for the content :)

[@UltraGameCoder]

you can also return a condition instead of if/else statement: return (d < distSqrt(x,y,px,py)); instead of if (d < distSqrt(x,y,px,py)) { return true; } else { return false; } same goes with simple values. let's assume we want to return 5 or 10. return ((d < distSqrt(x,y,px,py))? 5 : 10); this is how it works:if (condition)? if true : if false; condition = your check. ? = which side do I need to execute. left side of : symbol = if true is the result of the condition. right side of : symbol = if false is the result of the condition. hope you've all learned something by this message.

[@carlosaugustopinheirodesousa]

Fantastic teaching skills man! Great work!

[@TheCodingTrain]

So nice to hear, thank you!

[@NerdMusician]

Why bother paying Netflix when we have The Coding Train. Your tutorials have been so useful (and entertaining)!

[@FlavourlessLife]

Lol great tutorial, Blob Man. Not only am I learning lots, but you're also cheering me up in the process!

[@birendrayasas6764]

i wish he is our lecturer...so i wont skip classes.awesome teaching skills...

[@mikouh]

You should do a video on emotion detection with openCV, I currently have a project to do, and I wanted to do a smile detector and some typography animations, but I can't seem to find very many good stuff on this subject. Love your work btw

[@DogwafflDan]

I love that you're covering some fairly complex subjects. There are a few areas in computer graphics that continue to baffle me: Working in gradient domain and converting back to RGB color model. FFT for 2D images fluid simulation

[@benjaminrigsby1606]

I learned two things from this video. DIY blob detection and DIY v-neck. Thank you.

[@quakquak6141]

I think you could obtain a better result with a much lower treshold and a different way to calculate the distance between a pixel and a blob, you need to calculate the shortest distance between any pixel of the blob and the given pixel, this way if the red area is big you won't get a lot of small blobs next to one another even if you don't exclude small blobs

[@TheCodingTrain]

oh this is a great point! I'll try to make these improvements in a future session.

[@quakquak6141]

I'm glad that I was helpful, I noticed that if you had a treshold of 20 (let's say 20 pixel of distance using the dist function and not the distSq one) if a blob becomes bigger than 40x40 pixel even if a red pixel sit right next to it's border it will be excluded from the blob

[@RameenFallschirmjager]

damn! this channel is a virtual university! you handle such deep concept like nanny sings a lullaby! I name you the Aaron of the computer programmer! now I just need to find the Moses!

[@ahmadsebaq2244]

Man, you're awesome. I've learned and known away more about programming from your videos. Thanks a million ^^.

[@TheCodingTrain]

Thank you so much!

[@pulkitgupta2434]

Dude, you are the coolest guy on the internet

[@kangachiken4374]

Was this in a stream!?!?! Did i miss it!?!?! Thanks for these awesome videos Dan, I'm 16 and learning to code in general, but your videos are so energetic and motivating! Inspiring stuff keep it up!

[@andrewlozano88]

haha , he makes me crack up up sometimes . fun to learn with Dan!

[@TheCodingTrain]

Thanks for watching!

[@georgexu9591]

Great work! I've read all your books and now finally see your face~ XD

[@thankyouthankyou1172]

omg , I kept laughing and laughing. so interesting playing with the blobs

[@yoyomemory6825]

OMG, I never knew that your code training course also covered the image processing part.!!! That' s greaaat! Thank you.,

[@geoffwagner4935]

I just had nearly one of the scariest moments of my life, the Coding Train channel glitched and said Zero Playlists! "NooooooooooooooooooooooooooooooooOOOO!"

[@chrisgavin]

Loving these classes, thanks!

[@absolutedogy346]

I feel like I want you to be my best friend, most genuine and good teacher I have ever seen on youtube, reminds me of classic thenewboston when I first got into programming, you really have a special talent and I wish I could just sit and absorb your knowledge forever

[@andrewlozano88]

YESSS!!! AWESOME!! GOES PERFECT WITH MY PROJECT!!

[@TheCodingTrain]

so glad to hear!

[@hamzahassan8024]

i loveddd it

[@saeidyazdani]

best teacher ever

[@SachinBhangale]

Thank you, very useful fun learning video.

[@esmeraldaazucena3438]

very nice tutorial..I have a question sir, is it possible that I can get the Object's height and width if I pixilate the object in an image

[@kamanyuhuhu9985]

Can i get shape detection without using coloring filter ?

[@hamzahassan8024]

thank you so much for sharing really inspired me to code more

[@sarahkatecre8]

Would be interesting to experiment with the smaller threshold (as code is in 14.44 of video), when you 'comment out' the line of code that clears the blobs, you are able to track more than one colour when the mouse is clicked because it retains the previous colour as well as the new one.... Would it be possible to also remove blobs of colour being tracked with a right mouse button click or key pressed? Would also be interesting to find the average /center location of that specific colour - on the entire screen or within the larger blobs which you develop at the end of the video (and in the next couple videos in this series of course XD)

[@TheCodingTrain]

I like all these ideas! You could definitely keep an array of colors you are seeking and add / remove from it!

[@hcemreyilmaz]

How can we decide that, our blob is circle or not? Do you have a tricky idea for this purpose?

[@yuyinghan6317]

How can I set the colors that need to be tracked without clicking with the mouse?

[@selvapriya4115]

i try to track white pixel in a image, when the pixel moves,it merge with non moving white pixels.. how can i separate moving and non moving pixel in a image using opencv c++

[@grillodesignstudio7334]

Hi, it s possible to have a tutorial about tracking and draw on capture video? Help!

[@atrashdingdong8870]

You're awesome! Thank you for making these videos, it brings so much joy into my life!

[@kalvinkrishna6879]

nice tutorial sir, but i still not get any idea, how could you create blob object?? how could you detect Red color into blob object?

[@funwithariyan7347]

THANK YOUUUUUU (is it weird that im learning this in middle school without my teacher telling me to)

[@ChicagoBob123]

Thanks so much. Love this.

[@nicolassaganias5155]

Hi! How can I track blobs of a video file, instead from the camera?

[@chamarakv98madushanka41]

Awsome broh...

[@JaviepalFordring1]

O MY GUD!!! YOU'RE THE BEST!!!! and, what's kind of ide is that?!?!!

[@TheCodingTrain]

It's processing.org!

[@darylbantillo2810]

Is it applicable on android studio and opencv? sorry im a newbie...

[@ferdinandmarquita9259]

can you make a video of circle shaped blob?

[@micmicmw]

Wonderful video thank you =)

[@thearajr7583]

could you show me how to track color possition? thank you :)

[@AfifatulMukaroh]

Amazing tutorial!!!!! :)

[@kumareshr5975]

Give some tutorials on webVR and openCV

[@sarlupe]

I'd love to see how to generate a high res heatmap such as codeflow.org/entries/2013/feb/04/high-performance-js-heatmaps/ from a lower res data input.

[@alvarobyrne]

how about using some search algorithm: h t t p s : / / en.wikipedia.org / wiki / Category:Search_algorithms. would this help p5js being more performant? the thing is i don't know which search algorithm to use.

[@TheCodingTrain]

Thanks for the suggestion!

[@Demirci91]

thanks bro

[@dazzid]

novel price for Daniel!

[@011azr]

Dude, why are you always seems so bubbly and happy? Is it drugs or something? xD

[@preadatordetector]

You need a refactor tool there buddy?

[@BentleyGadget]

You're a goof! But this was good!

[@SiddharthMaurya7]

Awesome!! :D

[@MrSerrakos]

You are amazing !!!!!! I learned too many things on Java and Processing W I T H O U T book in just O N E week through your videos !!!!! Thank you thank you thank you P.S. I would like some help if you have some time. 1. How do I change your code to detect objects that appear only in an area 250X250 pixels around the center of the screen? ( I did it through a change here : for (int x = 0; x < video.width; x++ ) { for (int y = 0; y < video.height; y++ ) but there is an error about the size of this : color currentColor = video.pixels[loc]; 2. How do I change your code to track only the biggest object that is appearing in the screen ???? Thank you

[@DamianReloaded]

11:44 BUG!!!!!!!!!!!!!!!!!!!!!

[@TheCodingTrain]

ZOINKS

[@DamianReloaded]

XD

[@jashum8554]

Watching you mimic with no sound is rather intertaining