Тёмный

52 - What is GMM and how to use it for Image segmentation? 

DigitalSreeni
Подписаться 107 тыс.
Просмотров 34 тыс.
50% 1

Gaussian mixture model (GMM) is a type of clustering algorithm that falls under the umbrella of unsupervised machine learning techniques. As the name indicates, GMM models each cluster to a Gaussian distribution with specific mean and variance. It follows the expectation maximization algorithm principles to find the required parameters to model different Gaussians.
This tutorial explains the basics of GMM and also goes through the process of writing few lines of code to segment a microscope image.
The code from this video is available at: github.com/bns...

Опубликовано:

 

12 сен 2024

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 36   
@mcapro
@mcapro 3 года назад
Thanks for making this so simple - appreciate your effort in taking out time to explain in such clarity.
@DigitalSreeni
@DigitalSreeni 3 года назад
Thanks for watching!
@rongliao9255
@rongliao9255 3 года назад
This is amazing! Complex concepts are explained with great examples and illustrations!
@TwistedFate123
@TwistedFate123 3 года назад
I really like this type of explanations. Intuitive and not daunting. Keep it up!
@felip6180
@felip6180 4 года назад
once again... amazing job doing an intro into mixture models! tnahk you very much for this!
@DigitalSreeni
@DigitalSreeni 4 года назад
My pleasure!
@makarpronin2008
@makarpronin2008 4 месяца назад
Thank you! Really nice explanations.
@vdl1738
@vdl1738 3 года назад
One of the best videos I have ever seen. Thanks !!
@DigitalSreeni
@DigitalSreeni 3 года назад
So nice of you
@kundankumaragnuru6174
@kundankumaragnuru6174 Год назад
Hi, I am learning Machine learning and image processing concepts using your youtube tutorials and I am very glad I found your tutorials very useful and helpful. I have a query regarding the segmentation concept using GMM (unsupervised model). I have around 2 thousand images with no annotated data. I am trying to resolve segmentation problem using an unsupervised ML algorithm (GMM). 1. DO i have to fit and predict each image individually and get segmentation results ? 2, should i split the set of images into 2 sets (test folder and train folders) manually and fit the GMM model with the train folder and predict on the test folder ? [procedure you have explained in supervised machine learning]
@medhifamibelle1375
@medhifamibelle1375 2 года назад
thanks a lot, love the theory followed with the code, and the visual exapmles
@adoniskon1783
@adoniskon1783 3 года назад
you are the best teacher out there
@DigitalSreeni
@DigitalSreeni 3 года назад
Thank you.
@chinaguy101
@chinaguy101 2 года назад
very intuitive example video, very helpful!
@DigitalSreeni
@DigitalSreeni 2 года назад
Glad you think so!
@krishj8011
@krishj8011 3 года назад
Thanks..great video. Nice details and demonstrations..
@nickcorn93
@nickcorn93 3 года назад
This channel is really a treasure! Keep up the amazing work.
@shiyuanlin1047
@shiyuanlin1047 4 года назад
great videos!
@anthonyfix007
@anthonyfix007 4 года назад
thanks. you save my time
@user-tf6gv4nf3p
@user-tf6gv4nf3p 9 месяцев назад
hello sir, can you please help in resolving the problem of how to generate binary masks for day-3 human embryo and create masks for the image.
@umairsabir6686
@umairsabir6686 4 года назад
Hey @Sreeni, thanks for introducing me to this amazing technique which out stands many other clustering Algorithms. One question, everytime I run GMM clustering it gives me 3 different result each time and it keeps repeating these different results all the 3 times. There is one of the result in those 3 results that I am looking for. How can I select that particular result while running GMM ?
@DigitalSreeni
@DigitalSreeni 4 года назад
In my example I assigned my reshaped output to a parameter called 'segmented'. Since I provided n_components=4 I see values 0, 1, 2, and 3 in my output which I labeled segmented. This is just a numpy array so if you are interested in only one value then just set all those values to 1 and all others to 0. Basically you havea numpy array so it is up to your creativity and skill to extract the relevant information.
@mayureshnivatkar5374
@mayureshnivatkar5374 2 года назад
I performed the above code in google colab and I'm unable to adjust the brightness like you did so is there a way for me to change it , cause all I'm getting now is a black image just like you got.
@mohamedomar-rp3kz
@mohamedomar-rp3kz 2 года назад
Could I ask how to show the Gaussian distributions obtained by GMM
@zeeshanpatel6358
@zeeshanpatel6358 4 года назад
Hi, thank you for these tutorials. I was attempting to use the plant cell image you used in this tutorial, but it appears that it is not in your GitHub. Is it possible to send a link to the file? Thank you.
@DigitalSreeni
@DigitalSreeni 4 года назад
Some images are copy protected so I removed them from Github. You can use any image to test the code.
@md.alamintalukder3261
@md.alamintalukder3261 Год назад
Thanks a lot
@DigitalSreeni
@DigitalSreeni Год назад
No worries!
@aakashavashthi8930
@aakashavashthi8930 4 года назад
Hey .. i appreciate your work its awesome, I just have some queries that if i dont want to use image J for viewing my segmented image,what should i follow?
@DigitalSreeni
@DigitalSreeni 4 года назад
Try the multidimensional viewer on aPeer.com. It is a free (for academia and personal use) platform for image analysis and it also provides 500GB storage with embedded 3D viewer. Since it is browser based you don’t need any special hardware to view your images, just a chrome browser would do. Full disclosure, I work on the APEER team and it is designed for the research community. Check it out, you may like it. www.apeer.com
@aakashavashthi8930
@aakashavashthi8930 4 года назад
@@DigitalSreeni Sir, is there any chance to get it done using Cv library or something in code as i want to use this image to my further processing algorithm which is continous in nature.?
@khaledjbaili8654
@khaledjbaili8654 4 года назад
Thank you for the wonderful Tutorial. However, how to i threshold the image with out using image j. I mean I tried the code on my test image, it works good except that the image is black and i expect to get a masked image where only the red object is outlined in order to use it for detection. I also don't understand what happens by changing the maximum displayed value to 2 !1. Would appreciate it for some explanation
@DigitalSreeni
@DigitalSreeni 4 года назад
Please watch my videos 28 and 32.
@michaelakotodebrah8532
@michaelakotodebrah8532 4 года назад
Please can you help me with a code on Background subtraction using the GMM image segmentation
@DigitalSreeni
@DigitalSreeni 4 года назад
Not sure what you mean by background subtraction using GMM. You can use this approach to segment background. But if you are looking for background subtraction then search for rolling ball in my channel.
@user-ql5ex6kg4j
@user-ql5ex6kg4j 8 месяцев назад
This town is okay
Далее
58 - What are Gabor filters?
25:59
Просмотров 52 тыс.
Мужа или парня
00:42
Просмотров 14 тыс.
Clustering (4): Gaussian Mixture Models and EM
17:11
Просмотров 281 тыс.
Gaussian Mixture Model | Object Tracking
15:56
Просмотров 32 тыс.
Gaussian Mixture Model
15:07
Просмотров 105 тыс.
This is the Difference of Gaussians
19:03
Просмотров 252 тыс.
Gaussian Mixture Models
17:27
Просмотров 70 тыс.