I have a dataset of few thousand images and i need to count the number of cows within a particular region in an image. The problem is some of the images have only half a face, half a body or just a tail in the image. I am attaching some of such images with the cow i am interested to count bounded in red box. [IMG_0029.JPG](/upfiles/14634376083621543.jpg) [IMG_0050.JPG](/upfiles/14634376547887718.jpg) [IMG_0124.JPG](/upfiles/14634376862737428.jpg) [IMG_0016.JPG](/upfiles/14634377039416195.jpg)
Which algorithm would work the best to detect the cows in scene? I was thinking of using HAAR and training it with all possible results like half a face, face, body corner etc.
Is there any other better way?
Or does someone see a problem with this approach?
If you know someone who could help, please tag him/her
↧