How can we apply a few-shot using image with mask and text prompt

timojl / clipseg

This repository contains the code of the CVPR 2022 paper "Image Segmentation Using Text and Image Prompts".

Other

1.13k stars 107 forks source link

How can we apply a few-shot using image with mask and text prompt #22

Closed forever-rz closed 1 year ago

forever-rz commented 1 year ago

Thanks for your contribution,I'm still a bit confused on how to do a few- shot? Is it just a matter of continuing to train using the training code or or fine-tune?

timojl commented 1 year ago

For a few-shot setting, I recommend running the pre-trained model multiple times (i.e. once for every few-shot sample) and averaging the predictions. This way, you do not need to re-train.