一、数据集转化
python
import json
import os
from tqdm import tqdm
import glob
import os.path as osp
def json_to_txt(jsonfilePath, resultDirPath):
"""
jsonfilePath: labelme标注好的*.json文件所在文件夹
resultDirPath: 转换好后的*.txt保存文件夹
"""
class_names = []
jsonfileList = glob.glob(osp.join(jsonfilePath, "*.json"))
for jsonfile in tqdm(jsonfileList, desc='Processing'):
with open(jsonfile, "r", encoding='UTF-8') as f:
file_in = json.load(f)
shapes = file_in["shapes"]
txt_file = osp.join(resultDirPath, osp.basename(jsonfile).replace(".json", ".txt"))
with open(txt_file, "w") as file_handle:
for shape in shapes:
class_name = shape["label"]
if class_name not in class_names:
class_names.append(class_name)
class_id = class_names.index(class_name)
points = shape["points"]
normalized_points = [
f"{point[0] / file_in['imageWidth']} {point[1] / file_in['imageHeight']}"
for point in points
]
file_handle.write(f"{class_id} {' '.join(normalized_points)}\n")
with open(osp.join(resultDirPath, 'classes.txt'), 'w') as f:
f.write('\n'.join(class_names))
if __name__ == "__main__":
jsonfilePath = r"E:\241231数据扩充图片\标签" # 要转换的json文件所在目录
resultDirPath = r"D:\TXT" # 要生成的txt文件夹
json_to_txt(jsonfilePath=jsonfilePath, resultDirPath=resultDirPath)二、TXT标签验证
验证转化后的的标签是否正确，代码如下：
python
import os
import numpy as np
import cv2
IMG_FORMATS = ['.jpg', '.png', '.jpeg']
COLORS = [(0, 0, 0), (128, 0, 0), (0, 128, 0), (128, 128, 0), (0, 0, 128), (128, 0, 128), (0, 128, 128),
(128, 128, 128), (64, 0, 0), (192, 0, 0), (64, 128, 0), (192, 128, 0), (64, 0, 128), (192, 0, 128),
(64, 128, 128), (192, 128, 128), (0, 64, 0), (128, 64, 0), (0, 192, 0), (128, 192, 0), (0, 64, 128),
(128, 64, 12)]
def get_files(img_path, label_path):
"""
获取图像和标签文件列表
"""
img_list = [i for i in os.listdir(img_path) if os.path.splitext(i)[-1].lower() in IMG_FORMATS]
label_list = [i.replace('.txt', '') for i in os.listdir(label_path) if i.endswith('.txt')]
assert img_list, f"在 {img_path} 中未找到图像文件"
assert label_list, f"在 {label_path} 中未找到标签文件"
img_without_label = set(i.rsplit('.', 1)[0] for i in img_list) - set(label_list)
if img_without_label:
print(f"没有对应标签的图像: {list(img_without_label)}")
ok_img = list(set(i.rsplit('.', 1)[0] for i in img_list) - img_without_label)
return ok_img, os.path.splitext(img_list[0])[-1].lower()
def convert_and_draw(x, w, h, img):
"""
转换坐标并绘制多边形和标签
"""
label, *points = x
points = np.array(points, dtype=np.float32).reshape(-1, 2) * [w, h]
points = points.astype(np.int32)
color = COLORS[int(label) % len(COLORS)]
cv2.fillPoly(img, [points], color=color)
cv2.putText(img, text=str(int(label)), org=tuple(points[0]), fontFace=cv2.FONT_HERSHEY_COMPLEX,
fontScale=1, color=color, thickness=3)
return img
def main(img_path, label_path, outfile, if_show, show_num):
if os.path.isdir(img_path):
ok_img, img_ext = get_files(img_path, label_path)
for i, img_name in enumerate(ok_img, 1):
img = cv2.imread(os.path.join(img_path, img_name + img_ext))
h, w = img.shape[:2]
with open(os.path.join(label_path, img_name + '.txt'), 'r') as f:
labels = [x.split() for x in f.read().strip().splitlines()]
for label in labels:
img = convert_and_draw(label, w, h, img)
if if_show:
cv2.namedWindow('Image with label', cv2.WINDOW_NORMAL)
cv2.imshow('Image with label', img)
cv2.waitKey(0)
if i == show_num:
break
else:
img = cv2.imread(img_path)
h, w = img.shape[:2]
with open(label_path, 'r') as f:
labels = [x.split() for x in f.read().strip().splitlines()]
for label in labels:
img = convert_and_draw(label, w, h, img)
if if_show:
cv2.namedWindow('Image with label', cv2.WINDOW_NORMAL)
cv2.imshow('Image with label', img)
cv2.waitKey(0)
if __name__ == '__main__':
# 修改输入图片文件夹
img_path = r"C:\Users\Admin\Desktop\ultralytics-8.2.87\datasets\coco8-seg\images\train/000000000034.jpg"
# 修改输入标签文件夹
label_path = r"C:\Users\Admin\Desktop\ultralytics-8.2.87\datasets\coco8-seg\labels\train/000000000034.txt"
# 输出文件夹
outfile = './'
# 是否展示绘制的图片
if_show = True
# 最大展示图片的数量(按空格切换)
show_num = 3
main(img_path, label_path, outfile, if_show, show_num)三、结果显示
收工！！