利用log日志文件生成loss曲线_mmpose怎么将log

先贴一下自己的log文件的截图
这里以score_loss为例
代码如下

from matplotlib import rcParams
import matplotlib.pyplot as plt
import re


##显示中文
rcParams['font.family'] = 'sans-serif'
rcParams['font.sans-serif'] = 'SimSun,Times New Roman'

text = ''
file = open('log.txt')  ##我的log.txt文件是放在了和python文件同级别目录下
for line in file:
    text += line
file.close()
all_list = re.findall('loss: .*[0-9]',text)

train_loss = []
for i in all_list:
    train_loss.append(float(i.split('score_loss: ')[1].split(', grouperr:')[0]))  
     ##分割出score_loss的数值

plt.plot(train_loss,label='score_loss')

plt.legend(loc='best')
plt.show()

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26