2

I'd like to build tree crown detection software using OpenCV. Is it a good approach to do it using Haar-feature based Cascade Classifier? The tree crown I'd like to detect is oil palm trees. A sample of some images of the trees is below: palm trees

Is it doable if I use Haar Feature-based Cascade Classifier in OpenCV?

JohnAllen
  • 7,317
  • 9
  • 41
  • 65
jajamaharaja
  • 135
  • 14
  • What have you tried? How many positive images of these trees can you generate/acquire? Assuming you can acquire enough images of both these trees and `other trees that will also be in the area`, I believe you should be able to detect these. – JohnAllen Apr 05 '16 at 23:49
  • @JohnAllen I've tried to detect the crown using HSV thresholding and blob analysis, but it worked poorly when the images contain a lot of grass (which is have same color to the tree crown). The positive images i can acquire probably between 500 and 900 is it good enough ? – jajamaharaja Apr 10 '16 at 06:24
  • Is it good enough depends on how similar the other trees are to your type of tree. I would look at OpenCV's object detection. I wrote a tutorial here: johnallen.github.io/opencv-object-detection-tutorial – JohnAllen Apr 10 '16 at 20:01

0 Answers0