注意数据集中有很多增强图片，其中有47个类别(数字+大写字母+少部分小写)绝大部分均为一图多字符，剩余类别都是一个图一个字符。由于数字1和大写i在手写很难区分，还有大写C和小写C根本无法区分，数字0和大小写字母o，字母b和数字6在手写状态都很难区分开，可能容易误检测。标注时候还是区分标注的。

数据集格式：Pascal VOC格式+YOLO格式(不包含分割路径的txt文件，仅仅包含jpg图片以及对应的VOC格式xml文件和yolo格式txt文件)

图片数量(jpg文件个数)：38934

标注数量(xml文件个数)：38934

标注数量(txt文件个数)：38934

标注类别数：62

所在仓库：firc-dataset

标注类别名称(注意yolo格式类别顺序不和这个对应，而以labels文件夹classes.txt为准):["0","1","2","3","4","5","6","7","8","9","A","B","C","D","E","F","G","H","I","J","K","L","M","N","O","P","Q","R","S","T","U","V","W","X","Y","Z","a","b","c","d","e","f","g","h","i","j","k","l","m","n","o","p","q","r","s","t","u","v","w","x","y","z"]

每个类别标注的框数：

0 框数 = 3951

1 框数 = 3881

2 框数 = 3756

3 框数 = 3917

4 框数 = 3900

5 框数 = 3863

6 框数 = 3981

7 框数 = 4121

8 框数 = 3748

9 框数 = 3893

A 框数 = 4025

B 框数 = 4200

C 框数 = 3856

D 框数 = 3986

E 框数 = 3933

F 框数 = 3946

G 框数 = 4026

H 框数 = 3855

I 框数 = 3891

J 框数 = 3811

K 框数 = 3950

L 框数 = 4045

M 框数 = 4078

N 框数 = 3871

O 框数 = 3899

P 框数 = 3802

Q 框数 = 4033

R 框数 = 3933

S 框数 = 3926

T 框数 = 3926

U 框数 = 3888

V 框数 = 3894

W 框数 = 3872

X 框数 = 3784

Y 框数 = 4054

Z 框数 = 3888

a 框数 = 4006

b 框数 = 3874

c 框数 = 165

d 框数 = 3734

e 框数 = 3896

f 框数 = 4111

g 框数 = 4016

h 框数 = 4026

i 框数 = 308

j 框数 = 284

k 框数 = 171

l 框数 = 165

m 框数 = 165

n 框数 = 3888

o 框数 = 171

p 框数 = 165

q 框数 = 3913

r 框数 = 3896

s 框数 = 168

t 框数 = 4062

u 框数 = 166

v 框数 = 165

w 框数 = 166

x 框数 = 165

y 框数 = 165

z 框数 = 165

总框数：187559

使用标注工具：labelImg

标注规则：对类别进行画矩形框

重要说明：暂无

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，数据集只提供准确且合理标注

图片预览：

标注例子：