torch.histc

Running small sample segmentation ASGNet When , Calculation mIOU When , stay intersectionAndUnionGPU(output, target, K, ignore_index=255) in , Using the torch.histc Calculate the histogram ：

def intersectionAndUnionGPU(output, target, K, ignore_index=255):
    # 'K' classes, output and target sizes are N or N * L or N * H * W, each value in range 0 to K - 1.
    assert (output.dim() in [1, 2, 3])
    assert output.shape == target.shape
    output = output.view(-1)
    target = target.view(-1)
    output[target == ignore_index] = ignore_index
    intersection = output[output == target]
    area_intersection = torch.histc(intersection, bins=K, min=0, max=K-1)
    area_output = torch.histc(output, bins=K, min=0, max=K-1)
    area_target = torch.histc(target, bins=K, min=0, max=K-1)
    area_union = area_output + area_target - area_intersection
    return area_intersection, area_union, area_target

torch.histc Of API：

torch.histc(input, bins=100, min=0, max=0, *, out=None) → Tensor

Calculate the histogram of tensor .
Elements are classified as min and max Between cells of equal width . If min and max All zero , Then use the minimum and maximum values of the data .
Elements below the minimum and above the maximum will be ignored .
Parameters：

• input（Tensor）– Input $ZhangThe\; amount$.
• bins（int）– Number of histogram boxes
• min（int）– The lower limit of the range （ Include ）
• max（int）– The upper limit of the range （ Include ）

Returns

• output： Histogram with $ZhangThe\; amount$Tensor The output represents

Example ：

>>> torch.histc(torch.tensor([1., 2, 1]), bins=4, min=0, max=3)
 Output ：tensor([ 0.,  2.,  1.,  0.])

explain ： The histogram box is 0、1、2、3;
among

• 0 The quantity of is 0
• 1 The quantity of is 2
• 2 The quantity of is 1
• 3 The quantity of is 0

So the output tensor([0, 2, 1, 0])