33 - Grain size analysis in Python using watershed

[@sobitsingh8737]

How can i show in the image that which labels and area corresponds to what region in the image

[@rootkitraven]

great tutorial. but i have some questions.. i am a mineral processing engineer and often in the field we need to take a quick photo of rocks to evaluate our mining blasting or crushing performance and we are using split desktop software and we use a known scale. is this possible to put a scale beside sample and later in code we apply that scale (pixels to microns) and also return the size distribution graph? (using python) ...many thanks..

[@DigitalSreeni]

Place a known scale (like a ruler) beside your rock samples when taking the photo. In your Python code, you can manually select or automatically detect this scale. Calculate the pixels-to-microns ratio based on the known length of your scale. Then use this value to convert your pixel dimensions to microns or millimeters or other scales.

[@meghna3783]

Thank you so much and lots of love for these tutorial

[@DigitalSreeni]

My pleasure 😊

[@f1chtl]

@@DigitalSreeni Hello Sreeni! I am glad i found your Channel! There are so many useful ressources. I have one Question: Could i use some parts of your code for my BSc thesis? If yes, how should i cite you? Best regards!

[@pejmanyarahmadi349]

Hi dear Sreeni I have a question I already work on the stool sample images that take from the light microscope and it's very noisy that the noisy objects such as bacteria has overlap with the target objects and I can't segment the objects properly. What can I do for remove the noises from images without disfigure the target objects? can I send you a sample image? sorry if I didn't write my question very well, cause English isn't my primary language Thank you for your very useful tutorials

[@aldoblack5982]

Hi. Any chance to write the code on scikit-image? I am stuck on "conncectedComponents" function. I cannot seem to find it on scipy or scikit image. Thank you.

[@davidw.9711]

can you make a videonon the bodenmiller pipeline?

[@nikilottes9912]

how can i finde the file

[@aimagefrombydgoszcz]

Brilliant! Thank you for this lesson :)

[@radjivbewi1829]

Hi Sir, From France. First, thank a lot for your amazing videos and stips on python. That helps me a lot. I tried to study the grains soil distribution using the code you provide on this video but I don’t have contours around some grains even they are visibles, and so, because the are different colors than others (I think). Please, how can I do in order to get all grains no matter their color ? Thank in advance. Best regards!

[@sathishkumarpalani9511]

pixel to micron conversion for all image or it will change according to the image kindly explain sir

[@DigitalSreeni]

Pixel to micron conversion applies to the images you are analyzing using the code. If your images change in pixel size you need to change it in the code. You can also try to write code where it reads metadata and automatically updates the pixel size but that requires a bit of experience.

[@dancers-p1k]

Hi Srini, Thank you so much for the info. I'm trying to build a model to identify the grain/lentils (like Toor dal, chana Dal etc..), But I'm not successful. Could you please help me with which libraries will help me to do this? Thank you so much

[@DigitalSreeni]

The best approach would be multiclass classification using deep learning. May be my video 158 will help... kzbin.info/www/bejne/b3jdl4iwf8qsn5I

[@dancers-p1k]

@@DigitalSreeni thank you so much for the reply

[@paulocaldas8533]

Hey Sreeni! Great video! quick question, if you already have most of the info about each object when you get the markers with connectedComponents, why do you need the watershed function? I missed the point I guess... bc it seems that those final labels are the same than the markers you get before

[@DigitalSreeni]

Watershed is used to separate connected objects. If individual object separation is not of interest then you do not need to perform watershed.

[@sairaakram3196]

Thank you. I'm watching your videos for my learning and I got the purpose indeed. Can you please make video for the segmentation of a satellite like sentinel 2 image for classifying that image?

[@samarafroz9852]

Keep it up

[@barbituricacid]

"Let me not vocalize my every thought" :D :D :D

[@dubanvilladiego9629]

Thank you!!!

[@aminparvizi5112]

Nice :)