=================================================================================

A classic, most popular algorithm for image segmentation technique is called watershed. This is normally used for separating similar objects in the image that are touching each other. The process of watershed algorithm is:
         i) Create a border betweeb the overlapping objects in order to separate the touching objects.
         ii) Calculate a distance transform. This step takes binary images as inputs, and pixel intensities of the points inside the foreground regions are replaced by their distance to the nearest pixel with zero intensity (background pixel). This is done with the fucntion distance_transform_edt() from scipy library, which calculates the exact Euclidean distance transform.
         iii) Find local maxima points Binary image.
         iv) Label the marks.

SentenceTransformers, as shown in 4252, is a framework for state-of-the-art sentence, text and image embeddings in Python.

Figure 4252. SentenceTransformers.

============================================

Image Segmentation with watershed. code:
         
Input:         
         
Output: