运行环境vs code，代码如下：

import cv2
import math
import numpy as np
import os

def tile_cut(in_img_path,  out_img_path, valid_size, patch_size, is_GT):
    """
    滑动窗口裁剪
    Args:
        in_img_path(str): 输入大图路径.
        out_img_path (str): 输出的预测大图结果路径.
        valid_size (int): 预测小图的有效区域大小.
        patch_size (int): 预测小图的区域大小.
        is_GT (bool): 是否是真值文件.
    """
    img_file_name = os.path.basename(in_img_path)
    img_name = img_file_name[:-4]

    in_img = None
    if is_GT:
        in_img = cv2.imread(in_img_path, cv2.IMREAD_GRAYSCALE)
    else:
        in_img = cv2.imread(in_img_path)

    if in_img is None:
        print(in_img_path)

    #print(in_img.shape) #(行，列、通道数)

    # 确定渔网（256*256）左上角和右下角中心像素坐标
    quotient_cols = math.floor(in_img.shape[1] / valid_size) #向下取整
    residue_cols = in_img.shape[1] % valid_size #取余
    residue_cols = 1 if residue_cols > 0 else 0
    num_cols = quotient_cols + residue_cols

    quotient_rows = math.floor(in_img.shape[0] / valid_size) #向下取整
    residue_rows = in_img.shape[0] % valid_size #取余
    residue_rows = 1 if residue_rows > 0 else 0
    num_rows = quotient_rows + residue_rows

    center_minx = valid_size // 2
    center_miny = valid_size // 2
    center_maxx = valid_size // 2 + valid_size * (num_cols - 1)
    center_maxy = valid_size // 2 + valid_size * (num_rows - 1)

    # 根据渔网（256*256）左上角和右下角中心像素坐标确定需要填充的四至距离
    left_border = 0
    right_border = 0
    top_border = 0
    bottom_border = 0

    if (center_minx - patch_size // 2) < 0:
        left_border = patch_size // 2 - center_minx
    if (center_miny - patch_size // 2) < 0:
        top_border = patch_size // 2 - center_miny
    if (center_maxx + patch_size // 2) > in_img.shape[1]:
        right_border = center_maxx + patch_size // 2 - in_img.shape[1]
    if (center_maxy + patch_size // 2) > in_img.shape[0]:
        bottom_border = center_maxy + patch_size // 2 - in_img.shape[0]

    # 将四周填充行列
    in_img_pad = cv2.copyMakeBorder(in_img, top_border, bottom_border, left_border, right_border, cv2.BORDER_REFLECT_101)

    # 预测和拼接
    final_logit =  None
    for i in range(0, num_cols):
        for j in range(0, num_rows):

            # 填充图坐标系下的中心像素坐标
            center_x = valid_size // 2 + left_border + valid_size * i
            center_y = valid_size // 2 + top_border + valid_size * j

            # 判断中心像素坐标是否有误
            if center_x - valid_size < 0:
                print("center_x - valid_size < 0")
                return None
            if center_y - valid_size < 0:
                print("center_y - valid_size < 0")
                return None
            if center_x + valid_size > in_img_pad.shape[1]:
                print("center_x + valid_size > in_img_pad.shape[1]")
                return None
            if center_y + valid_size > in_img_pad.shape[0]:
                print("center_y + valid_size > in_img_pad.shape[0]")
                return None

            # 预测图在填充图坐标系下的位置，并抠取需要预测的小图
            left_top_x = center_x - valid_size
            left_top_y = center_y - valid_size
            right_bottom_x = center_x + valid_size
            right_bottom_y = center_y + valid_size
            cropped = in_img_pad[left_top_y:right_bottom_y, left_top_x:right_bottom_x]
            cv2.imwrite(out_img_path + img_name + '_' + str(center_x) + '_' +  str(center_x) + '.png', cropped)


if __name__ == '__main__':

    out_img_path = ' ' 
    in_img_dir = ' '

    valid_size = 256
    patch_size = 512

    is_GT = False

    in_img_list = os.listdir(in_img_dir)
    for in_img_file in in_img_list:
        if in_img_file[-3:] == 'png':
                print(in_img_file)
                im_img_path = in_img_dir + in_img_file
                tile_cut(im_img_path, out_img_path, valid_size, patch_size, is_GT)