[paddleseg source code reading] paddleseg calculation dice

$Dice$ Coefficient is a set similarity measure function , It is usually used to calculate the similarity of two sets , The value range is $[0,1]$:

$$Dice(X,Y)=\frac{2|X\cap Y|}{|X|+|Y|}$$

$|.|$ Represents a function that takes the number of set elements

$$Dice(X,Y)=\frac{twopersonhand\; overSetelementplainOfindividualCount\times 2}{twopersonelementplainCountThe\; amountAndand}$$

PaddleSeg In the implementation of ：

def dice(intersect_area, pred_area, label_area):
    """ Calculate DICE.  Calculation  DICE  value . Args: intersect_area (Tensor): The intersection area of prediction and ground truth on all classes. pred_area (Tensor): The prediction area on all classes. label_area (Tensor): The ground truth area on all classes. Returns: np.ndarray: DICE on all classes. float: mean DICE of all classes. """
    #  Transform the three into  np.adarray
    intersect_area = intersect_area.numpy()
    pred_area = pred_area.numpy()
    label_area = label_area.numpy()
    
    #  The number of two set elements 
    union = pred_area + label_area
    class_dice = []
    for i in range(len(intersect_area)): #  Start iterating in each class 
        if union[i] == 0:
        	#  If the molecule is  0, That is, the current image has no such elements 
            dice = 0
        else:
            dice = (2 * intersect_area[i]) / union[i]
        class_dice.append(dice)
        
    mdice = np.mean(class_dice)
    return np.array(class_dice), mdice

intersect_area, pred_area, label_area These three parameters are passed :

calculate_area(pred, label, num_classes, ignore_index=255)

Calculated , Follow up with a blog ：
[PaddleSeg Source code reading ] PaddleSeg Calculation mIoU