您好，我运行您的PPOCRLabel的文件转换成yolo格式，运行之后找不到生成后的文件噢？请问是我哪里步骤有误吗？
import os
import cv2
filename = "/home/lixinru/PaddleOCR/dataset/LABEL/Label.txt" #需要转换的标注文件
"""
# 读取数据文件
"""
def read_txt(file_name):
data = []
file = open(file_name, 'r', encoding='utf-8')
file_data = file.readlines() # 读取所有行
for row in file_data:
tmp_list = row.split(' ')
data.append(tmp_list) # 将每行数据插入data中
file.close()
return data
"""
# 创建文件夹
"""
def create_folder(filename):
filename = filename.strip()
filename = filename.rstrip("\\")
isExists = os.path.exists(filename)
if not isExists:
os.makedirs(filename)
print(filename + "创建成功")
return True
else:
print(filename + "已存在")
return False
"""
# 创建txt标注文件
"""
def txt_create(name, msg):
desktop_path = filename # 新创建的txt文件的存放路径
full_path = desktop_path + "/" + name.split('.')[0] + '.txt'
if os.path.exists(full_path):
file = open(full_path, 'a', encoding='utf-8')
file.write(msg + '\n')
file.close()
else:
file = open(full_path, 'w', encoding='utf-8')
file.write(msg + '\n')
file.close()
"""
# 坐标归一化
"""
def normalization(xmin, ymin, xmax, ymax, img_w, img_h):
x = round((xmin + xmax) / (2.0 * img_w), 6)
y = round((ymin + ymax) / (2.0 * img_h), 6)
w = round((xmax - xmin) / (1.0 * img_w), 6)
h = round((ymax - ymin) / (1.0 * img_h), 6)
return x, y, w, h
"""
# 每张图片的尺寸
"""
def get_img_size(img_path):
print(img_path)
mat = cv2.imread(img_path)
if mat is None:
return mat
return mat.shape
if __name__ == "__main__":
data = read_txt("/home/lixinru/PaddleOCR/dataset/LABEL/Label.txt") # 原始数据表 #需要转换的标注文件
create_folder("/home/lixinru/PaddleOCR/dataset/LABEL/turning")#新建的转换后文件路径
for img_data in data:
img_name = str(img_data).split("\\t")[0].split('/')[1] # 每行数据提取图片名
xywh = str(img_data).split("\\t")[1].split("\\n")[0].replace("',", "").replace("'", "").strip('[').strip(
']').replace("false", "1").replace("true", "1") # 每行数据提取坐标信息
# print(xywh)
xywh = eval(xywh) # 转换为元组或者字典
img_size = get_img_size(img_name)
if img_size is None:
continue
if type(xywh) == tuple:
for xy_line in xywh:
yolo_data = normalization(xy_line['points'][0][0], xy_line['points'][0][1], xy_line['points'][2][0],
xy_line['points'][2][1], img_size[1],
img_size[0]) # 处理每个字典的坐标信息，转换为归一化后的yolo标注格式
# print("0 %s %s %s %s" % (str(yolo_data[0]) , str(yolo_data[1]) , str(yolo_data[2]) , str(yolo_data[3])))
# print(img_name)
txt_create(img_name, "0 %s %s %s %s" % (
str(yolo_data[0]), str(yolo_data[1]), str(yolo_data[2]), str(yolo_data[3])))
if type(xywh) == dict:
yolo_data = normalization(xywh['points'][0][0], xywh['points'][0][1], xywh['points'][2][0],
xywh['points'][2][1], img_size[1], img_size[0])
txt_create(img_name,
"0 %s %s %s %s" % (str(yolo_data[0]), str(yolo_data[1]), str(yolo_data[2]), str(yolo_data[3])))
# print("0 %s %s %s %s" % (str(yolo_data[0]) , str(yolo_data[1]) , str(yolo_data[2]) , str(yolo_data[3])))
# print(img_name)
在/home/lixinru/PaddleOCR/dataset/LABEL/turning目录下找找看有没有。如果没有的话可能是你代码中设置filename变量的时候有问题，你可以在代码中的txt_create函数中添加一些调试信息，把一些参数变量都print（）打印出来看一下是具体哪步有问题。
如果以上回答对您有所帮助，点击一下采纳该答案～谢谢