Context-self contrastive pretraining for crop type semantic segmentation.

Michail Tarasiou Riza Alp Güler Stefanos Zafeiriou

Published in: CoRR (2021)

Keyphrases

semantic segmentation
superpixels
street scenes
conditional random fields
weakly supervised
object class
scene classification
ground truth
small number
input image
object detection
graph structure
biologically inspired
label transfer