steps.sam2.model.trainer

trainer

Functions:

Name	Description
train	Training step for fine-tuning a SAM2 model on a custom dataset.

train(model, dataset_collection, sam2_repo_path)

Training step for fine-tuning a SAM2 model on a custom dataset.

This step prepares the dataset and masks, configures the environment, and launches the SAM2 training loop.

Parameters:

Name	Type	Description	Default
model	Model	The Picsellia model instance containing pretrained weights and logging context.	required
dataset_collection	DatasetCollection[CocoDataset]	The dataset collection for training, expected to include a 'train' split.	required
sam2_repo_path	str	Path to the local SAM2 repository used for training.	required