数据集格式：labelme格式(不包含mask文件，仅仅包含jpg图片和对应的json文件)
图片数量(jpg文件个数)：642
标注数量(json文件个数)：642
标注类别数：18
标注类别名称:["csat_ct","csv_ct","tcrung_ct","kneo_ct","LD110kV-QTPC","dcleo_ct","cdien_polyme_ct_ban","ddan_ct","leo_ct","dcset_ct","kdo_ct","ddan_tt_tua","ddan_ct_tua","ddan_ct_vatla","cdien_ttinh_ct_ban","cdien_polyme_ct_rach","mnoi_ct","cdien_ttinh_ct_vo"]
每个类别标注的框数：
csat_ct (电缆接头) count = 112
csv_ct (电缆终端) count = 76
tcrung_ct (变压器) count = 101
kneo_ct (绝缘子破损) count = 80
LD110kV-QTPC (110kV电力设备) count = 1
dcleo_ct (电杆倾斜) count = 11
cdien_polyme_ct_ban (电缆聚合物绝缘故障) count = 71
ddan_ct (电杆本体) count = 178
leo_ct (避雷器) count = 16
dcset_ct (电杆基础) count = 26
kdo_ct (绝缘子污秽) count = 33
ddan_tt_tua (电杆拉线) count = 4
ddan_ct_tua (电杆拉线) count = 46
ddan_ct_vatla (电杆附属物) count = 60
cdien_ttinh_ct_ban (电力设备绝缘故障) count = 90
cdien_polyme_ct_rach (电缆聚合物断裂) count = 61
mnoi_ct (互感器) count = 14
cdien_ttinh_ct_vo (电力设备外壳破损) count = 70
使用标注工具：labelme=5.5.0
所在仓库：firc-dataset
标注规则：对类别进行画多边形框polygon
图片分辨率：640x640
重要说明：可以将数据集用labelme打开编辑，json数据集需自己转成mask或者yolo格式或者coco格式作语义分割或者实例分割
特别声明：本数据集不对训练的模型或者权重文件精度作任何保证
图片预览：
标注例子：
