Training Set (large images)
small image version
In the results, the images are as follows:
- original: Original
- seg: Segmentation using Felzenszwalb & Huttenlocher algorithm
- cluster: for each cluster of segments with more than 1 element, that cluster
is colored a random color (segments in singleton clusters made greyscale) - pop: original image with segments in singleton clusters made greyscale.
original |
seg |
cluster |
pop |
bike-racks-and-windows |
|||
![]() |
![]() |
![]() |
|
bollards1 |
|||
![]() |
![]() |
![]() |
|
dandelions |
|||
![]() |
![]() |
![]() |
|
flowers |
|||
![]() |
![]() |
![]() |
|
upson-floor-dots1 |
|||
![]() |
![]() |
![]() |
|














