Dear OpenEDS organizers,
As in the rules of SS challenge
Entrants may not manually create additional semantic segmentation masks other than what is already given in the Data Set to train the Model.
Can we use a deep network to generate the masks for these images and use in the training, Or, we can only use the images with providing masks?
Thank you very much.
Hi there,
It would be okay to utilized some tech to generate some pseudo masks for those unlabeled images, as long as they are not manually annotated.
Please also keep in mind that in order to be considered as a winner, participants would need to share the code for the creation of the additional semantic segmentation masks and the whole process should be replicable on our end.
Please let us know for further questions.