以数据集BSDS500为例,将HED-BSDS/train目录下的相应的图片与边缘提取后的图像一一对应,写入lst文件中。
目的:为之后训练网络提供前提条件
下述为HED-BSDS/train/aug_data_scale_0.5与HED-BSDS/train/aug_gt_scale_0.5文件夹下的图像对应生成在C:/Users/Liu/Desktop/DataSets目录下的output2.lst文件。
import os
def create_lst_file(root_folder, output_file):
with open(output_file, 'w') as lst_file:
for subfolder in os.listdir(root_folder + "/aug_data_scale_0.5"):
# 获取对应的aug_data和aug_gt文件夹路径
aug_data_folder = os.path.join(root_folder, "aug_data_scale_0.5", subfolder)
aug_gt_folder = os.path.join(root_folder, "aug_gt_scale_0.5", subfolder)
# 获取aug_data文件夹中的所有文件
aug_data_files = [os.path.join(aug_data_folder, filename) for filename in os.listdir(aug_data_folder)]
# 获取aug_gt文件夹中对应的文件
aug_gt_files = [os.path.join(aug_gt_folder, filename) for filename in os.listdir(aug_gt_folder)]
# 确保两个文件夹中的文件数量相同
if len(aug_data_files) != len(aug_gt_files):
raise ValueError(f"The number of files in {aug_data_folder} and {aug_gt_folder} must be the same.")
# 将文件路径对应起来并写入.lst文件
for data_file, gt_file in zip(aug_data_files, aug_gt_files):
data_file = data_file.replace("\\", "/") # 将反斜杠替换为正斜杠
gt_file = gt_file.replace("\\", "/") # 将反斜杠替换为正斜杠
data_file = data_file.replace("C:/Users/Liu/Desktop/DataSets/", '') # 去掉前缀部分
gt_file = gt_file.replace("C:/Users/Liu/Desktop/DataSets/", '') # 去掉前缀部分
lst_file.write(f"{data_file} {gt_file}\n")
if __name__ == "__main__":
root_folder = "C:/Users/Liu/Desktop/DataSets/HED-BSDS/train" # 根文件夹的路径
output_file = "C:/Users/Liu/Desktop/DataSets/output2.lst" # 输出的.lst文件名
create_lst_file(root_folder, output_file)
生成结果如下: