279 - An introduction to object segmentation using StarDist library in Python

For details on installing StarDist library and other examples:
github.com/stardist/stardist
Stardist helps with object detection of star-convex shapes, both in 2D and 3D.
Code generated in the video can be downloaded from here:
github.com/bnsreenu/python_for_microscopists/blob/master/279_An_introduction_to_object_segmentation_using_StarDist.ipynb
All other code:
github.com/bnsreenu/python_for_microscopists

[@ahmedraafat8769]

Thanks Dr. Sreenivas. I learnt a lot from you and still learning

[@anubratadas1]

Thanks a lot Sreenivas! your videos are very helpful

[@rahulrajendran7392]

I am woking with brightfield multiplex images similar to immuno flourecence, hope it will be useful for my research, waiting for your next video!

[@rahulrajendran7392]

you are saving my internship!!!!!!

[@DigitalSreeni]

I'm glad I could help.

[@tapansharma460]

great video

[@solomonadeyemi53]

Thanks alot for this amazing working ......Please, what will you suggest I use for object segmentation (e.g buildings, crop field etc ) from satellite imagery ?

[@leo.neumann]

Amazing video! Thank you!... What should I use if I want to segment non star convex objects?

[@zmjjiang2766]

Thanks for your explanation, sir. i am looking forward to your next video. Hoping to know how to obtain the edge or contour of every object after we detected it and how to save them in file.

[@willberger96]

Amazing!

[@rahulds001]

I hope this comment reaches you. We the remote sensing students are clueless when it comes to prepare our satellite data into patches and masks to be used for segmentation.All video works on already prepared datsets. I kindly request you to look into the matter and come out with a video. We are and will be very grateful. Thank you. PS- it will benefit in landslide mapping by segmenting the landslides as it is very difficult to visually mark the landsldies within an area.

[@hridaanshugusain3102]

Hi, I was just wondering if the labels for the ground truth have to be multi-class labels or can it be single class? I previously used label-studio and labeled my slides with a single class, but when I import the ground truth into the stardist, it never works. Any idea what the issue is? Thanks

[@takfarines5522]

Hi sir, I have a question, I am trying to read and write a raster with gdal library but I can save only one band, the output is a panchromatique image I want to keep all the bands. Any ideas?

[@ranimsaidi9564]

Hello, great content thank you! can you please tell me in which video you talked about add the border in each class and doing the training for the U-Net ?

[@DigitalSreeni]

kzbin.info/www/bejne/f3jReoWmasmKlcU

[@annicetrazafindratovolahy1512]

Hi, can you make a video explaining how to use the swin transformer in TensorFlow as an autoencoder architecture? Thank you

[@bineethavijayan4244]

usefull video sir. Will you be able to do videos on audio dataset too

[@jeanbon3173]

great great great

[@devrimonder1304]

Could you give me the video reference which explains your proposed improvement strategy for u-net connected object problem ? Thanks

[@AKASHKUMAR-jz9yv]

Sir I am working on microstructure segmentation of multiphase alloy in which grain boundaries are not clearly visible instead a diffused interface is there and there is a lot of colour contrast within the microstructure. I am unable to use Machine learning for segmentation. Please make a video related to my problem.

[@lewismackay7612]

I can't seem to find the link to the google colab or any of the documentation in the link you provided...am I looking in the right place? Great video by the way.

[@gerokatseros]

You are my favourite teacher! Can we use it to detect crops on sattelite images as long as they are normal polygons?

[@DigitalSreeni]

Technically yes but I cannot guess without trying. StarDist helps in segmenting objects that are 'roundish' or even polygons with no weird shapes. So it may work on crops. I never tried it so no words of wisdom based on personal experience.

[@subirrakshit1563]

Python pretrained model for biological image.

[@y.shrestha6936]

Thanks!

[@DigitalSreeni]

Thank you very much for your kindness Shrestha. Keep learning.

[@gabrielmonacoribeirodasilv8643]

do you think this would work segmenting and quantifying rock pores in SEM images?

[@DigitalSreeni]

Yes. You may have to invert the LUT but I am sure it works for pore segmentation.

[@lalitsingh5150]

Sir, please make a video on superpixel segmentation and extraction of specific superpixels....thanks in advance

[@cplusplus-python]

Great, as always.! Thanks, Prof. Any chance about Mask-RCNN? multi-object segmentation?

[@DigitalSreeni]

Yes, mask-RCNN videos are coming.

[@cplusplus-python]

@@DigitalSreeni Thank you 👏🏻👍🏻🙏🏻

[@RafiKhalil]

@@DigitalSreeni Respected Sir everyone is waiting for Mask-RCNN videos.!!

[@eriklogan3996]

😉 𝙥𝙧𝙤𝙢𝙤𝙨𝙢

[@lalitsingh5150]

Thanks!

[@DigitalSreeni]

Thank you very much.