Here is a picture of a complex cell adhesion. I tried to segment it using the watershed algorithm, but it didn't work. Can someone give me some pointers?My sincerest thanks！

Or, what other algorithms can solve such complex glued pictures?

from skimage.feature import peak_local_max
from skimage.morphology import watershed
from scipy import ndimage
import numpy as np
import cv2

input_img1 = cv2.imread('F:\\twf\\code\\Image enhancement\\image1\\24.png')
input_img = cv2.resize(input_img1, (0,0),fx=0.25,fy=1/3,interpolation=cv2.INTER_CUBIC)
cv2.imshow("input_img",input_img)
input_imgsp=input_img.copy()
gray = cv2.cvtColor(input_img, cv2.COLOR_BGR2GRAY)
adaptivethresh = cv2.adaptiveThreshold(gray , 255,
                                     cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV,25,5)
cv2.imshow("adaptivethresh ",adaptivethresh)

#Binarization

black = np.ones([512, 512, 3], np.uint8)*255
blackb=np.ones([512, 512, 3], np.uint8)*255
__,conts, hier = cv2.findContours(adaptivethresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
l = 14.00
a = 10.00
for i, contour in enumerate(conts):
    cnt = contour
    cntArea = cv2.contourArea(cnt)
    len = cv2.arcLength(cnt, True)
    if cntArea <a and len < l:
        cv2.drawContours(black, conts, i, (0, 0, 255), 1)
cv2.imshow('black',black)
blackg = cv2.cvtColor(black, cv2.COLOR_BGR2GRAY)
ret2, threshblack = cv2.threshold(blackg,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
kernel = np.ones((2, 2), np.uint8)
result2=adaptivethresh-threshblack
cv2.imshow('result2',result2)

#remove noise

sure_bg = cv2.dilate(result2,kernel,iterations=1)
D = ndimage.distance_transform_edt(result2)
localMax = peak_local_max(D, indices=False, min_distance=5,labels=result2)
markers = ndimage.label(localMax, structure=np.ones((3, 3)))[0]
labels = watershed(-D, markers, mask=result2)
output = np.zeros((input_imgsp.shape[0],input_imgsp.shape[1], 3), np.uint8)
mask = np.zeros(gray.shape, dtype="uint8")
n=0
mask = np.zeros(gray.shape, dtype="uint8")
for label in np.unique(labels):
    if label == 0:
        continue
    mask[labels == label] = 255

cv2.imshow("mask", mask)
#watershed segmentation of adherent spores

num_labels, lab, stats, centroids = cv2.connectedComponentsWithStats(mask , connectivity=8)
output = np.zeros((input_imgsp.shape[0],input_imgsp.shape[1], 3), np.uint8)
for i in range(1, num_labels):
    mask = lab == i
    output[:, :, 0][mask] = np.random.randint(0, 255)
    output[:, :, 1][mask] = np.random.randint(0, 255)
    output[:, :, 2][mask] = np.random.randint(0, 255)
cv2.namedWindow("Output", 0)
cv2.imshow('Output', output)

#Assign different colors to different connected domains
#But the result failed, and the segmentation was not successful
cv2.waitKey()

enter image description here

enter image description here