统计数据集标签中目标数量

1
import os
2

3
def count_non_empty_lines_in_folder(folder_path):
4
    total_lines = 0
5

6
    # 遍历文件夹下的所有文件
7
    for root, dirs, files in os.walk(folder_path):
8
        for file in files:
9
            if file.endswith('.txt'):
10
                file_path = os.path.join(root, file)
11
                with open(file_path, 'r', encoding='utf-8') as f:
12
                    # 统计非空行
13
                    non_empty_lines = [line for line in f if line.strip() != '']
14
                    total_lines += len(non_empty_lines)
15

16
    return total_lines
17

18
# 定义train和test文件夹的路径
19
train_folder = 'labels/train'
20
test_folder = 'labels/test'
21

22
# 统计每个文件夹中非空行的数量
23
train_lines = count_non_empty_lines_in_folder(train_folder)
24
test_lines = count_non_empty_lines_in_folder(test_folder)
25

26
print(f"Train文件夹中目标的总数: {train_lines}")
27
print(f"Test文件夹中目标的总数: {test_lines}")