2D IOU与3D IOU计算

在目标检测任务中，常用IOU作为评价指标，它的含义就是交并比，如下图：

计算方式很直观，将预测框与ground truth框之间的交集比上两者的并集。

代码如下：

def iou_2d(box1,box2):
    '''
        box [x1,y1,x2,y2]   分别是两对角定点的坐标
    '''
    area1 = (box1[2]-box1[0])*(box1[3]-box1[1])
    area2 = (box2[2]-box2[0])*(box2[3]-box2[1])
    area_sum = area1 + area2
    
    #计算重叠部分 设重叠box坐标为 [x1,y1,x2,y2]
    x1 = max(box1[0], box2[0])
    y1 = max(box1[1], box2[1])
    x2 = min(box1[2], box2[2])
    y2 = min(box1[3], box2[3])
    if x1 >= x2 or y1 >= y2:
        return 0
    else:
        inter_area = (x2-x1)*(y2-y1)
    return inter_area/(area_sum-inter_area)

def iou_3d(box1, box2):
    '''
        box [x1,y1,z1,x2,y2,z2]   分别是两对角定点的坐标
    '''
    area1 = (box1[3]-box1[0])*(box1[4]-box1[1])*(box1[5]-box1[2])
    area2 = (box2[3]-box2[0])*(box2[4]-box2[1])*(box2[5]-box2[2])
    area_sum = area1 + area2

    #计算重叠部分 设重叠box坐标为 [x1,y1,z1,x2,y2,z2]
    x1 = max(box1[0], box2[0])
    y1 = max(box1[1], box2[1])
    z1 = max(box1[2], box2[2])
    x2 = min(box1[3], box2[3])
    y2 = min(box1[4], box2[4])
    z2 = min(box1[5], box2[5])
    if x1 >= x2 or y1 >= y2 or z1 >= z2:
        return 0
    else:
        inter_area = (x2-x1)*(y2-y1)*(z2-z1)

    return inter_area/(area_sum-inter_area)

2D和3D代码几乎完全一样，都是先算两个框各自面积/体积，再算重叠部分，最后计算IOU。