288 - Nuclei segmentation using StarDist and tracking using Trackpy in python
ฝัง
- เผยแพร่เมื่อ 16 ต.ค. 2024
- Code generated in the video can be downloaded from here:
github.com/bns...
All other code:
github.com/bns...
This video walks you through the process of segmenting nuclei in fluorescence microscope images using StarDist (pre-trained deep learning model), followed by tracking them using trackpy.
Trackpy installation: soft-matter.git...
Trackpy strongly recommends using Conda and not pip. Therefore, we need to set up conda environment for our colab notebook.
This specific topic has been covered in a separate tutorial: • Installing Conda in Go...
Just making a habit of liking your video at the very start as i always forget to like your video because of the details you fill in it. Thanks for the content sir
Thank you for the effort to prepare such great material for your ppt every week. I've been learning a lot watching your videos and I am getting more interested in semantic segmentation models.
I would really appreciate it if you can prepare or give us a kick idea or code about unsupervised semantic segmentation, please. Thank you in advance!
Another excellent presentation! I would not have known about these libraries on my own. Another set of tools to add to the toolbox. Stardist would have probably outperformed my half baked Hough transform based algorithm on a previous project.
Thank you very much for your generosity. And yes, there are a lot of libraries that are useful and actively maintained. I learn about them from talking to many people in the image processing community.
Another great lesson! Your channel is one of those things that make internet worth. Thanks! Im currently in a project using label studio for multiclass segmantion and Im having a hard time on how could I stack the masks of a given image into a single png. Do you have any specifics on that in your channel (couldn't find) or to provide any tip? I think data preparation is one of the most important and complex part of the process.
Read one mask at a time for each image, change pixel values for a given mask to a specific number, do this for all masks representing all classes, save the numpy array into a single png. I am sure I have done this in one of my videos but I haven't done an exclusive video on merging individual masks into a single multiclass mask.
sir please make a video about nnunet usage for costum dataset on colab..
@DisitalSreeni Could you please reply. I have a humble request. If I want to segment whole images (not single image) using starDIST of a dataset what should I do? Similarly, for tracking as well. Plz help. It's urgent.
Hello Sir.
Is there any problem if i use 16 bit mask with 8 bit image for unet training purpose?
No
sir can u make tutorial for classification with 3D image and multiclass. i look the people make classification 3D images use binner so in here i hope u can help me about this.
Thankyou so much for this presentation. Sir, please make a video on HoverNet for nuclei segmentation and classification.
Thanks for the great videos you make. Could you please make a video about Self Supervised Learning (SSL) & Representation Learning and/or transformers as well?
thank you sir, this is so interesting, can i use this one for fish movement in the tank sir?
is stardist working with 3D dataset , it will be highly appreciated if you show us how tomuse it with 3d dataset
StarDist github has good examples for 3D: github.com/stardist/stardist/tree/master/examples/3D
Is the images data also available?
Can you make videos on unsupervised semantic segmentation. It would be greatly appreciated. Thankyou
importing stardist's libraries makes kernel restart (jupyter)! I have stuck in this step
it seems the problem is with tensorflow, the error is zsh: illegal hardware instruction ipython (mac M2)
Hello Sir, please accept my request on LinkedIn. I am getting one error while implementing your code.