Fix Imports in Livecell Inference

doc/finetuned_models.md

@@ -26,8 +26,8 @@ See for example the [2d annotator example](https://github.com/computational-cell
 
 As a rule of thumb:
 - Use the `_lm` models for segmenting cells or nuclei in light microscopy.
-- Use the `_em` models for segmenting ceells or neurites in electron microscopy.
-    - Note that this model does not work well for segmenting mitochondria or other organelles becuase it is biased towards segmenting the full cell / cellular compartment.
+- Use the `_em` models for segmenting cells or neurites in electron microscopy.
+    - Note that this model does not work well for segmenting mitochondria or other organelles because it is biased towards segmenting the full cell / cellular compartment.
 - For other cases use the default models.
 
 See also the figures above for examples where the finetuned models work better than the vanilla models.

doc/python_library.md

@@ -6,8 +6,8 @@ import micro_sam
 ```
 
 The library
-- implements function to apply Segment Anything to 2d and 3d data more conviently in `micro_sam.prompt_based_segmentation`.
-- provides more and imporoved automatic instance segmentation functionality in `micro_sam.instance_segmentation`.
+- implements function to apply Segment Anything to 2d and 3d data more conveniently in `micro_sam.prompt_based_segmentation`.
+- provides more and improved automatic instance segmentation functionality in `micro_sam.instance_segmentation`.
 - implements training functionality that can be used for finetuning on your own data in `micro_sam.training`.
 - provides functionality for quantitative and qualitative evaluation of Segment Anything models in `micro_sam.evaluation`.
 

finetuning/.gitignore

@@ -2,3 +2,4 @@ checkpoints/
 logs/
 sam_embeddings/
 results/
+*.sh

finetuning/livecell_finetuning.py

@@ -22,10 +22,10 @@ def get_dataloaders(patch_shape, data_path, cell_type=None):
     """
     label_transform = torch_em.transform.label.label_consecutive  # to ensure consecutive IDs
     train_loader = get_livecell_loader(path=data_path, patch_shape=patch_shape, split="train", batch_size=2,
-                                       num_workers=8, cell_types=cell_type, download=True,
+                                       num_workers=16, cell_types=cell_type, download=True,
                                        label_transform=label_transform, shuffle=True)
     val_loader = get_livecell_loader(path=data_path, patch_shape=patch_shape, split="val", batch_size=1,
-                                     num_workers=8, cell_types=cell_type, download=True,
+                                     num_workers=16, cell_types=cell_type, download=True,
                                      label_transform=label_transform, shuffle=True)
     return train_loader, val_loader
 

micro_sam/evaluation/livecell.py

@@ -15,7 +15,7 @@
 from segment_anything import SamPredictor
 from tqdm import tqdm
 
-from ..instance_segmentation import AutomaticMaskGenerator, EmbeddingMaskGenerator
+from ..instance_segmentation import AutomaticMaskGenerator, _EmbeddingMaskGenerator
 from . import automatic_mask_generation, inference, evaluation
 from .experiments import default_experiment_settings, full_experiment_settings
 
@@ -169,7 +169,7 @@ def run_livecell_amg(
 
     if use_mws:
         amg_prefix = "amg_mws"
-        AMG = EmbeddingMaskGenerator
+        AMG = _EmbeddingMaskGenerator
     else:
         amg_prefix = "amg"
         AMG = AutomaticMaskGenerator
@@ -231,8 +231,8 @@ def run_livecell_inference() -> None:
     # - automatic mask generation (auto)
     # if none of the two are active then the prompt setting arguments will be parsed
     # and used to run inference for a single prompt setting
-    parser.add_argument("-f", "--full_experiment", action="store_true")
     parser.add_argument("-d", "--default_experiment", action="store_true")
+    parser.add_argument("-f", "--full_experiment", action="store_true")
     parser.add_argument("-a", "--auto_mask_generation", action="store_true")
 
     # the prompt settings for an individual inference run
@@ -242,7 +242,7 @@ def run_livecell_inference() -> None:
     parser.add_argument("-n", "--negative", type=int, default=0, help="No. of negative prompts")
 
     # optional external prompt folder
-    parser.add_argument("--prompt_folder", help="")
+    parser.add_argument("--prompt_folder", help="Provide the path where all input point prompts will be stored")
 
     args = parser.parse_args()
     if sum([args.full_experiment, args.default_experiment, args.auto_mask_generation]) > 2: