简单的python性能测试

最近入门语义分割，生成数据集的时候非常慢，觉得自己肯定是代码写的不够好，于是决定简单的测试一下性能，看看哪个函数可以优化

import cv2
import cProfile
from codes.utils import *

def main():
    in_size = 572
    out_size = 388
    image_name = 'data/pic/high_quality_architectural_6066_.png'
    msk_name = 'data/label/high_quality_architectural_6066_.png'
    img = cv2.imread(image_name) #镀金来的数据是BGR模式的
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # [height, width, channel]
    img_as_np = np.asarray(img)
    img_as_np = img_as_np.transpose((2, 1, 0)) # [channel, width, height]

    msk = cv2.imread(msk_name) #镀金来的数据是BGR模式的
    msk = cv2.cvtColor(msk, cv2.COLOR_BGR2RGB) # [height, width, channel]
    msk_as_np = np.asarray(msk)
    msk_as_np = msk_as_np.transpose((2, 1, 0)) # [channel, width, height]

    img_as_np = expand_img(img_as_np, 512) # 扩充到比standard要大，这样才可以裁剪
    msk_as_np = expand_img(msk_a