Integrating SSL into SegmentationTask #1937
Unanswered
Dodekaphonia
asked this question in
Q&A
Replies: 1 comment 1 reply
-
I feel like @isaaccorley would know a good way to do this. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello!
I'm currently working on a model for coastline change detection along the Antarctic coastline using Sentinel-1 images.
I followed @calebrob6's tutorial on LEVIR+ and tested this approach on my relatively small dataset of labeled changes, utilizing the
ResNet50_Weights.SENTINEL1_ALL_MOCO
weights. The results appeared promising.Considering further improvements, I considered the idea of pre-training a self-supervised learning (SSL) model and then using it as a backbone for the
SegmentationTask
. To stay within one framework, I attempted to utilize the BYOLTask, MoCoTask, and SimCLRTask. However, I encountered difficulties in finding a straightforward approach to achieve pass the '.ckpt' file into a SegmentationSo, my question is whether it is possible to pre-train one of the SSL tasks and then integrate it into the Segmentation process. Or should I explore alternative approaches?
Beta Was this translation helpful? Give feedback.
All reactions