computer vision - Applying K-Means to HOG-Descriptors -


i've extracted hog-descriptors using dlib , stored them in array of 2 dimensional arrays. now, want cluster descriptors using k-means.

how 1 perform such clustering? can think of 2 ways implement it:

  1. performing clustering line-wise. hence, 2 dimensional array separately.

  2. performing clustering lines @ once.

dlib contains program, imglab (in tools/imglab folder) has --cluster option. option k-means clustering on hog images. using angular distance metric particularly effective clustering hog vectors. more usual euclidean k-means.

so recommend using angular distance metric. can refer above tool/dlib specifics.


Comments

Popular posts from this blog

python - Selenium remoteWebDriver (& SauceLabs) Firefox moseMoveTo action exception -

html - How to custom Bootstrap grid height? -

transpose - Maple isnt executing function but prints function term -