Little python script to analyse output of a net
from validation import Validation
validation = Validation('path_df/file.csv', 'path_image_folder')
validation.report('path_save/name.pdf', 'My personnel report on Mnist')
To get a pdf file not too heavy, you can control the number of exemple taken :
from validation import Validation
validation = Validation('path_df/file.csv', 'path_image_folder')
validation.report('path_save/name.pdf', 'My personnel report on Mnist', nb_sample = 150)
If you want to not have the same exemple sampled you can add :
from validation import Validation
validation = Validation('path_df/file.csv', 'path_image_folder')
validation.report('path_save/name.pdf', 'My personnel report on Mnist', nb_sample = 150, randomize = True)
| file | y_true | class_1 | class_2 | class_3 | |
|---|---|---|---|---|---|
| 0 | class_1.png | 1 | 0.84 | 0.10 | 0.06 |
| 1 | class_2.png | 2 | 0.75 | 0.05 | 0.20 |
| 2 | class_3.png | 3 | 0.10 | 0.10 | 0.80 |
If not precised, file name column label MUST start with 'file'. Else, you can precise it by giving the list of file labels with the argument x_name
| file_0 | file_1 | y_true | class_1 | class_2 | class_3 | |
|---|---|---|---|---|---|---|
| 0 | class_1_x20.png | class_1_x10.png | 1 | 0.84 | 0.10 | 0.06 |
| 1 | class_2_x20.png | class_2_x10.png | 2 | 0.75 | 0.05 | 0.20 |
| 2 | class_3_x20.png | class_3_x10.png | 3 | 0.10 | 0.10 | 0.80 |
Or
from validation import Validation
validation = Validation('path_df/file.csv', 'path_image_folder', x_name = ['file_0', 'file_1'])
validation.report('path_save/name.pdf', 'My personnel report on Mnist')
| WARNING: Classes order is important |
|---|