watershed | Python implementation of the watershed image segmentation | Computer Vision library

In matlab, I can use these functions from the image processing toolbox in order to "clean" a distance transform and locate the seeds for watershed algorithm. The tutorials I found for python include thresholding the distance transform to get the seeds, which doesnt work well in my case because not all my features are the same size, and the intensity of the distance transform depends on absolute distance from the background.

I have attached an example of what those functions accomplish in matlab, where the white regions are the seeds. Is there a way I can achieve this in python, either with existing functions or coding it on my own?

I'm trying to use Skimage to segment an image with watershed, but I always get this error. Do you have a solution please?

AttributeError: module 'skimage.morphology' has no attribute 'watershed'

Source code : https://scikit-image.org/docs/0.12.x/auto_examples/xx_applications/plot_coins_segmentation.html

In the image I would like to have only the lungs in black not the background. The background [Top black and bottom black area of the image] must be white not black. How can I do that in python?. Code that resuls in the image above from an original grayscale image ("image") is:

I started app project, for image processing, using OpenCv 4.5.3 and Swift ( with C++ ). I'm fighting with watershaded alg. for a really long time... And i have no clue what did i do wrong. Just don't know...

Error :

I have an M-length array of 3-dimensional coordinate points as floats. I would like to create a 3-dimensional numpy array of a predefined shape filled with ellipsoids of a given float radius centered on these points. Because this is intended for image manipulation, I call each value in the array a "pixel". If these ellipsoids overlap, I would like to assign the pixel to the closer center by euclidean distance. The final output will be a numpy array with a background of 0 and pixels within ellipsoids numbered 1, 2,... M, as corresponding to the initial list of coordinates, similar to the output of scipy's ndimage.label(...).

Below, I have a naive approach which considers every position in the output array and compares it to every defined center, creating a binary array with a value of 1 for any pixel inside any ellipsoid. It then uses scikit-image to watershed this binary array. While this code works, it is prohibitively slow for my use, both because it considers every pixel and center pair and because it performs watershedding seperately. How can I speed up this code?

There is an R package in development that I would like to use called streamstats. What it does is delineate a watershed (within the USA) for a latitude & longitude point along a body of water and provides watershed characteristics such as drainage area and proportions of various land covers. What I would like to do is extract some watershed characteristics of interest from a data frame of several lat & long positions.

I can get the package to do what I want for one point

I'm still pretty new within the image-segmentation / OpenCV scene and hope you can help me out. Currently, I'm trying to calculate the percentage of the 2 liquids within this photo

It should be something like this (as an example)

I thought opencv watershed could help me but I'm unable to get it right. I'm trying to set the markers manually but I get the following error: (-215:Assertion failed) src.type() == CV_8UC3 && dst.type() == CV_32SC1 in function 'cv::watershed' (probably I got my markers all wrong)

If anyone can help me (maybe there is a better way to do this), I would greatly appreciate it

This is the code I use:

I have a table of metadata about attachments. Sometimes the auto PDF conversion does not work and it does not create a PNG file. I want a query I can run to see which PDF files this is happening on so I can go fix it..

In this case I want the results:

This has worked fine for months, but throws errors now. It's a pretty simple code: open Chrome and pull an element out. It's now telling me that it's unable to locate the element.

I've inspected the element with the Chrome window open and verify the XPATH and it matches to what is in the code exactly. I've taken a screenshot of the page it's running on and it all looks fine.

What am I missing?