segment-geospatial
segment-geospatial copied to clipboard
opengeos
Feature Request: Fine-tuning
Description
I know this is a big feature request, but the addition of fine-tuning support would be nice as a component of this package.
TorchGeo could probably be used nicely for sampling from large images when training
For now, we can only fine-tune what SAM allows. See the parameters below. It might take sometime before we can support other model fine-tuning. Contributions are always welcome.
https://github.com/opengeos/segment-geospatial/blob/main/samgeo/samgeo.py#L43
points_per_side: Optional[int] = 32,
points_per_batch: int = 64,
pred_iou_thresh: float = 0.88,
stability_score_thresh: float = 0.95,
stability_score_offset: float = 1.0,
box_nms_thresh: float = 0.7,
crop_n_layers: int = 0,
crop_nms_thresh: float = 0.7,
crop_overlap_ratio: float = 512 / 1500,
crop_n_points_downscale_factor: int = 1,
point_grids: Optional[List[np.ndarray]] = None,
min_mask_region_area: int = 0,
output_mode: str = "binary_mask",
Additional resources for fine-tuning
- https://encord.com/blog/learn-how-to-fine-tune-the-segment-anything-model-sam/
- https://github.com/ziqi-jin/finetune-anything
- https://github.com/luca-medeiros/lightning-sam
- https://github.com/tianrun-chen/SAM-Adapter-PyTorch
- https://github.com/facebookresearch/segment-anything/issues/5
- https://colab.research.google.com/drive/1F6uRommb3GswcRlPZWpkAQRMVNdVH7Ww?usp=sharing
- https://github.com/ArminMoghimi/Fine-tune-the-Segment-Anything-Model-SAM-
Hi @giswqs , any updates on how we can fine-tune on new custom classes for GroundingDINO, or maybe replacing GroundingDINO with other pre-trained models that produces bboxes for downstream SAM? Thanks in advance!
Is it possible to fine tune sam_geo with our own pertained model I have a dataset in form of yolov8 annotations can I use it with sam geospatial ?
