Few-Shot Fine-tuning

[PixxelAnit]

Hey Team,

Great work! Huge advancements can be seen from SAM to SAM3 in last 2.5 years. I just had a query:

Can you please provide some reference scripts to fine-tune the model for a different domain for few-shot segmentation? I can't find anything in the repo related to it. So, it will be a great help if you can provide any help regarding to it.

Thanks,
Anit

[alcinos]

Could you let us know what questions you still have after reading https://github.com/facebookresearch/sam3/blob/main/README_TRAIN.md?

[Huster-Hq]

Could you let us know what questions you still have after reading https://github.com/facebookresearch/sam3/blob/main/README_TRAIN.md?

Are the two datasets listed in the (https://github.com/facebookresearch/sam3/blob/main/README_TRAIN.md) both used for training SAM3 in image-level tasks?

[alcinos]

Roboflow100 and odinW are not part of the training mix of the released checkpoints. We report in the paper the few-shot results on those datasets (in that case, there is a checkpoint for each subset). We do not release those checkpoints

[DabeetDas]

What is the required format to finetune on a custom dataset with text inputs?

[cuppersd]

What is the required format to finetune on a custom dataset with text inputs?

[BahMoh]

I am new to these sort of LVLMs. Is there a way to fine-tune the model with image-text pairs?