kolomomo / labelme2yolo

labelme标签转换为yolo标签

labelme2yolo

labelme标签转换为yolo标签

python labelme2yolo(...)

参数：

labelme2yolo(labelme_image_path='labelme 图像文件夹目录',
                 labelme_json_path='labelme json文件夹目录',
                 save_data_path='转换yolo格式保存目录',
                 obj_name='提取的类别名称, labelme 里的 labels.txt',
                 dataset_type='train', # train/val/test
                 image_type='.jpg', # 图像格式后缀
                 )

labelme2yolo's Issues

labelme2yolo

labelme标注的是分割 json文件里面是一个个封闭的多边形及其类别？对吗？
下面的函数就是 打开一个图片的标注？一个个读取多边形，把多边形换成矩形方框用于检测？

with open(path_txt, 'w+') as ftxt:
data = json.load(fp)
self.height = data['imageHeight']
self.width = data['imageWidth']
for shape in data['shapes']:
label = shape['label']
shape_type = shape['shape_type']

                if shape_type != "rectangle":
                    print('只支持retangle标注')
                    break

                if label in self.labels:
                    idx = self.labels.index(label)
                    points = shape['points']
                    x_center, y_center, w, h = self.point_to_box(points)