Constantin Pape @cppape.bsky.social · Feb 4 Our latest preprint PEFT-SAM is on arxiv! We study parameter efficient finetuning to adapt SAM to biomedical images and introduce a new workflow for adaptation based on only two labeled images. See our workflow and improvements in the image, check out arxiv.org/abs/2502.00418 or read on for more.
11 likes 1 replies
?
Replies
We compare different parameter efficient finetuning methods, including LoRA, freezing model parts and QLoRA, for which we provide one of the first implementations for computer vision. The methods are tested for a diverse range of medical and microscopy segmentation tasks.