o .wi@sVddlmZmZddlZddlmZddlmZddlmZddl m Z ddl m Z ddl mZ d$d ed edeeddeddeeedfddf ddZ d%deded ed ededdeeeeff ddZ  d&dededeeddeedeeedfdef dd Z ! d'deded ed edeeddeddefd"d#ZdS)()OptionalUnionN)Tensor)Literal)_ignore_background)rank_zero_warn)_check_same_shape) _safe_dividemicroone-hot? num_classesinclude_backgroundaverager macroweightednone input_formatr index zero_divide)warnnanreturncCst|tr |dkrtd|dt|tstd|dgd}|dur5||vr5td|d|d|d vrAtd |d|d vrMtd |ddS) z%Validate the arguments of the metric.rzDExpected argument `num_classes` must be a positive integer, but got .zBExpected argument `include_background` must be a boolean, but got rNz)Expected argument `average` to be one of z or None, but got rzJExpected argument `input_format` to be one of 'one-hot', 'index', but got )r rrzNExpected argument `zero_divide` to be one of 1.0, 0.0, 'warn', 'nan', but got ) isinstanceint ValueErrorbool)r rrrrallowed_averager"f/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/torchmetrics/functional/segmentation/dice.py_dice_score_validate_argss  r$predstargetc Cst|||dkr#tjjj||ddd}tjjj||ddd}|jdkr1td|jd|s:t||\}}t t d|j}tj |||d }tj ||d }tj ||d }d|} ||} |} | | | fS) z8Update the state with the current prediction and target.r)r zJExpected both `preds` and `target` to have at least 3 dimensions, but got rdim) rtorchnn functionalone_hotmovedimndimrrlistrangesum) r%r&r rr reduce_axis intersection target_sumpred_sum numerator denominatorsupportr"r"r#_dice_score_update1s   r=r:r;r< zero_divisioncCst|dkrt|dkrtjd|jtjdS|dkr*tj|dd}tj|dd}t|||d}|dkr>tj|dd}|S|d kr\|d ur\t|tj|dd d |d}tj||dd}|S) z:Compute the Dice score from the numerator and denominator.rr)devicedtyper r'r+)r>rrNT)r,keepdim)r-alltensorr?floatr5r mean)r:r;rr<r>diceweightsr"r"r#_dice_score_computePs rHTc CsF|dkr tdtt||||t|||||\}}}t||||dS)a Compute the Dice score for semantic segmentation. Args: preds: Predictions from model target: Ground truth values num_classes: Number of classes include_background: Whether to include the background class in the computation average: The method to average the dice score. Options are ``"micro"``, ``"macro"``, ``"weighted"``, ``"none"`` or ``None``. This determines how to average the dice score across different classes. input_format: What kind of input the function receives. Choose between ``"one-hot"`` for one-hot encoded tensors or ``"index"`` for index tensors Returns: The Dice score. Example (with one-hot encoded tensors): >>> from torch import randint >>> from torchmetrics.functional.segmentation import dice_score >>> preds = randint(0, 2, (4, 5, 16, 16)) # 4 samples, 5 classes, 16x16 prediction >>> target = randint(0, 2, (4, 5, 16, 16)) # 4 samples, 5 classes, 16x16 target >>> # dice score micro averaged over all classes >>> dice_score(preds, target, num_classes=5, average="micro") tensor([0.4842, 0.4968, 0.5053, 0.4902]) >>> # dice score per sample and class >>> dice_score(preds, target, num_classes=5, average="none") tensor([[0.4724, 0.5185, 0.4710, 0.5062, 0.4500], [0.4571, 0.4980, 0.5191, 0.4380, 0.5649], [0.5428, 0.4904, 0.5358, 0.4830, 0.4724], [0.4715, 0.4925, 0.4797, 0.5267, 0.4788]]) Example (with index tensors): >>> from torch import randint >>> from torchmetrics.functional.segmentation import dice_score >>> preds = randint(0, 5, (4, 16, 16)) # 4 samples, 5 classes, 16x16 prediction >>> target = randint(0, 5, (4, 16, 16)) # 4 samples, 5 classes, 16x16 target >>> # dice score micro averaged over all classes >>> dice_score(preds, target, num_classes=5, average="micro", input_format="index") tensor([0.2031, 0.1914, 0.2500, 0.2266]) >>> # dice score per sample and class >>> dice_score(preds, target, num_classes=5, average="none", input_format="index") tensor([[0.1714, 0.2500, 0.1304, 0.2524, 0.2069], [0.1837, 0.2162, 0.0962, 0.2692, 0.1895], [0.3866, 0.1348, 0.2526, 0.2301, 0.2083], [0.1978, 0.2804, 0.1714, 0.1915, 0.2783]]) r zdice_score metric currently defaults to `average=micro`, but will change to`average=macro` in the v1.9 release. If you've explicitly set this parameter, you can ignore this warning.)r<)r UserWarningr$r=rH) r%r&r rrrr:r;r<r"r"r# dice_scorehs6rJ)r r r )r )r Nr )Tr r )typingrrr-rtyping_extensionsr*torchmetrics.functional.segmentation.utilsrtorchmetrics.utilitiesrtorchmetrics.utilities.checksrtorchmetrics.utilities.computer rr rDr$tupler=rHrJr"r"r"r#s           "