ziqi-jin / finetune-anything

Fine-tune SAM (Segment Anything Model) for computer vision tasks such as semantic segmentation, matting, detection ... in specific scenarios
MIT License
791 stars 56 forks source link

After the fine-tuning is SamAutomaticMaskGenerator applicable #30

Closed MustafaArikan closed 1 year ago

MustafaArikan commented 1 year ago

Hi,

very interesting work. After fine-tuning is the SamAutomaticMaskGenerator applicable to generate masks similar to the original SAM.

thx

ziqi-jin commented 1 year ago

The mask of SAM is an instance segmentation mask without category mark. The instance segmentation task that will be supported later will output the same instance segmentation mask as SAM, and there will be a separate field to store category information. For other tasks, such as matting, semantic segmentation, etc., the mask required by the corresponding task will be output