Adding a border around Region of Interest using python - python-3.x

Question

I have a code which takes images from a folder, crops the region of interest around it using the ROI function, and then removes the background using the rembg library. But I want a border around that image, around that specific object itself, like the one we get in segmentation, except keeping the colours and the object intact. [NOT A RECTANGULAR BORDER]. Can anyone help and tell me how to do it?
Here is my code for reference:
import cv2
import numpy as np
import os
from os.path import join
from os import listdir
from PIL import Image
from rembg import remove
path = 'Some path'
folder = 'Some other path'
count = 1
def image_resize(image, width = None, height = None, inter = cv2.INTER_AREA):
dim = None
(h, w) = image.shape[:2]
if width is None and height is None:
return image
if width is None:
r = height / float(h)
dim = (int(w * r), height)
else:
r = width / float(w)
dim = (width, int(h * r))
resized = cv2.resize(image, dim, interpolation = inter)
return resized
for filename in os.listdir(folder):
img = cv2.imread(os.path.join(folder,filename))
if img is not None:
img = image_resize(img, height = 600)
roi = cv2.selectROI(img)
print(roi)
im_cropped = img[int(roi[1]):int(roi[1]+roi[3]),int(roi[0]):int(roi[0]+roi[2])]
rs = str(count)
rem = remove(im_cropped)
cv2.imshow("Removed Image", rem)
cv2.imwrite(os.path.join(path, rs + '.jpg'), rem)
count = count + 1
cv2.waitKey(0)

Answer 1

As i assume you mean a simple outline, i would suggest the following:
import cv2
import numpy as np
# Threshold the image
thresh = img > threshold
# Use binary dilation to widen the area
thresh_dil = cv2.dilate(thresh, np.ones((3, 3)) , iterations=1)
# Get the outlines by substracting the dilated image with the original area
outlines = thresh_dil - thresh
# Superimpose the outlines on your original image
img_with_outlines = img.copy()
img_with_outlines[outlines > 0] = 255
This should draw a white line around your detected object.
Note: This approach works grayscale images. For full colour images you can apply it for each channel individually.