0

I am trying to separate a cow from a depth image.
enter image description here

When I use contours it separates most of it but fails to separate the fence when the cow is leaning on it. (Note: it is ok that the head is being removed from the cow, the application I am using it on works better if the head is removed) enter image description here

Here is the code I use to detect and remove contours. My idea is to remove them by size. It works when the cow is not touching the fence but in this case, doesn't work.

# Remove stuructures connected to the image border------------------------------------------------
# find contours in the image and initialize the mask that will be
cnts = cv2.findContours(BW3.copy(), cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
cnts = imutils.grab_contours(cnts)
mask = np.ones( BW3.shape[:2], dtype="uint8") * 255
# loop over the contours
for c in cnts:
    if cv2.contourArea(c) > 250000:
    # if the contour is bad, draw it on the mask
        cv2.drawContours(mask, [c], -1, 0, -1)
        BW3 = cv2.bitwise_and( BW3,  BW3, mask=mask)
    if cv2.contourArea(c) < 10000:
        cv2.drawContours(mask, [c], -1, 0, -1)
        BW3 = cv2.bitwise_and( BW3,  BW3, mask=mask)
        
cv2.imshow('H_Black and white',  BW3)
cv2.waitKey()

Is there any way to remove the fencing around the cow when it is touching? I have tried using HoughLinesP() with no luck, I am new to OpenCV so I could be going about it the wrong way. Another potential solution would be to crop the image, but the camera is in a slightly different location each time so cropping would have to be adjusted for each camera variation. Any advice would be appreciated.

Thank you

EDIT: The purpose of separating the cow from the background is to use volumetric estimation to determine the weight of the animal. If effectively implemented this will be a cheaper solution than a standard scale. This is for a research project (The project will be open-sourced, not monetized).

The original input depth image is cropped before any other code is run (All images reflect this except for the first depth image in this post) To get the contours I change the depth picture to HSV. Then take the Hue image and change that to black and white before running cv2.findCountours

HSV Black and White from hue

Here is a reconstruction of the depth values from the jet colormap (inverted values so it's easier to read visually):

enter image description here

Christoph Rackwitz
  • 11,317
  • 4
  • 27
  • 36
Robert Kadlec
  • 23
  • 5
  • 1
    please provide the depth image itself, in grayscale/metric, not false color. -- also explain the goal https://xyproblem.info/, **why** are you trying to get a cow contour? **what for**? – Christoph Rackwitz Dec 31 '21 at 17:02
  • 2
    take a depth image without a cow, when you want to find a cow subtract the empty image (keep pixel if closer), what is left is the cow – rioV8 Dec 31 '21 at 17:12
  • @ChristophRackwitz I have updated the question please lmk if I should include any other information. – Robert Kadlec Dec 31 '21 at 17:21
  • @rioV8 That is a great idea! I will try it out. thank you – Robert Kadlec Dec 31 '21 at 17:22
  • 1
    I remember a question like this. I also remember I said I'd approach this as a "background segmentation" task. take a picture of the empty stall as a baseline. in any picture, if a pixel's depth is close to baseline, it's background. one could even apply opencv's background subtraction code (there's a module in the main repo and one in the contrib repo) and have this adjust to any changes of the camera. – Christoph Rackwitz Dec 31 '21 at 20:24
  • 1
    the grayscale picture you provided isn't depth values. I have no idea how you made this, but it's not data. the cow's back looks concave in that version, which can't be right. -- congrats. you made me investigate how matplotlib defines colormaps and how to invert colormapped data (`jet`, FlannBasedMatcher). I've added a grayscale depth picture to your post. I'm sure I'll be using this in the future, so it's a useful bit of code to have. – Christoph Rackwitz Dec 31 '21 at 23:39

0 Answers0