优化: 更简单的评价函数计算方法

[BetterWorld-Liuser]

第一次PR, 如果有什么不对的地方还请原谅

主要是将评价函数修改成更简单的形式.

if __name__ == '__main__':
    from PIL import Image
    l, p = np.random.randint(0,3,size=(500,100)),np.random.randint(0,3,size=(500,100))
    print(get_miou_recall_precision(l, p, [0, 1, 2]))
    print(l_get_miou_recall_precision(l, p, [0, 1, 2]))

可通过上述代码进行正确性的验证.

(0.19795873925264684, {0: 0.3282282282282282, 1: 0.32790837854128996, 2: 0.3353221957040573}, {0: 0.3276378896882494, 1: 0.32860163092721234, 2: 0.33522218908440204})
(0.19795873925264684, {0: 0.3282282282282282, 1: 0.32790837854128996, 2: 0.3353221957040573}, {0: 0.3276378896882494, 1: 0.32860163092721234, 2: 0.33522218908440204})

输出一致.