saigege / darknet-loss-drawer

This repository is developed on the basis of pyqt5, mainly through the log files generated during the running of Darknet, and then draw loss files.

darknet-loss-Drawer

This repository is developed on the basis of pyqt5, mainly through the log files generated during the running of Darknet, and then draw loss files.

the release version can be used only in Windows

when you run you darknet using scipts below:

./darknet detector train cfg/coco.data cfg/yolov3.cfg

The output will like this:

Resizing

320

Loaded: 0.502709 seconds

Region 30 Avg IOU: 0.453259, Class: 0.406288, Obj: 0.616954, No Obj: 0.530078, .5R: 0.500000, .75R: 0.333333, count: 6

Region 42 Avg IOU: 0.237628, Class: 0.556286, Obj: 0.263673, No Obj: 0.461042, .5R: 0.142857, .75R: 0.000000, count: 7

Region 54 Avg IOU: 0.244585, Class: 0.550149, Obj: 0.428019, No Obj: 0.469195, .5R: 0.074074, .75R: 0.037037, count: 27

Region 30 Avg IOU: 0.306451, Class: 0.559028, Obj: 0.413168, No Obj: 0.530804, .5R: 0.000000, .75R: 0.000000, count: 4

Region 42 Avg IOU: 0.262898, Class: 0.576616, Obj: 0.381279, No Obj: 0.460506, .5R: 0.111111, .75R: 0.000000, count: 9

Region 54 Avg IOU: 0.255984, Class: 0.518751, Obj: 0.423396, No Obj: 0.469569, .5R: 0.074074, .75R: 0.000000, count: 27

Region 30 Avg IOU: 0.324626, Class: 0.472797, Obj: 0.618393, No Obj: 0.529930, .5R: 0.285714, .75R: 0.000000, count: 7

Region 42 Avg IOU: 0.289221, Class: 0.723829, Obj: 0.541236, No Obj: 0.462390, .5R: 0.166667, .75R: 0.000000, count: 6

Region 54 Avg IOU: 0.237150, Class: 0.541031, Obj: 0.491277, No Obj: 0.469132, .5R: 0.045455, .75R: 0.000000, count: 22

Region 30 Avg IOU: 0.358145, Class: 0.339489, Obj: 0.425689, No Obj: 0.529440, .5R: 0.333333, .75R: 0.333333, count: 3

Region 42 Avg IOU: 0.391667, Class: 0.668793, Obj: 0.368548, No Obj: 0.461217, .5R: 0.428571, .75R: 0.000000, count: 7

Region 54 Avg IOU: 0.214031, Class: 0.543846, Obj: 0.334361, No Obj: 0.469448, .5R: 0.000000, .75R: 0.000000, count: 29

1: 562.584595, 562.584595 avg, 0.000000 rate, 0.553693 seconds, 32 images

Loaded: 0.936840 seconds

Region 30 Avg IOU: 0.210395, Class: 0.535263, Obj: 0.585981, No Obj: 0.530203, .5R: 0.000000, .75R: 0.000000, count: 5

Region 42 Avg IOU: 0.246853, Class: 0.507559, Obj: 0.499927, No Obj: 0.459885, .5R: 0.125000, .75R: 0.000000, count: 8

Region 54 Avg IOU: 0.155895, Class: 0.499203, Obj: 0.295966, No Obj: 0.468601, .5R: 0.047619, .75R: 0.000000, count: 42

Region 30 Avg IOU: 0.343129, Class: 0.493684, Obj: 0.606848, No Obj: 0.528431, .5R: 0.200000, .75R: 0.200000, count: 5

Region 42 Avg IOU: 0.282754, Class: 0.626201, Obj: 0.483792, No Obj: 0.459979, .5R: 0.000000, .75R: 0.000000, count: 8

Region 54 Avg IOU: 0.205155, Class: 0.482449, Obj: 0.359561, No Obj: 0.469626, .5R: 0.027778, .75R: 0.000000, count: 36

Region 30 Avg IOU: 0.218516, Class: 0.643676, Obj: 0.513226, No Obj: 0.530422, .5R: 0.000000, .75R: 0.000000, count: 5

Region 42 Avg IOU: 0.223242, Class: 0.580035, Obj: 0.488847, No Obj: 0.460264, .5R: 0.111111, .75R: 0.000000, count: 9

Region 54 Avg IOU: 0.151966, Class: 0.455435, Obj: 0.314113, No Obj: 0.468514, .5R: 0.052632, .75R: 0.000000, count: 38

Region 30 Avg IOU: 0.362651, Class: 0.503255, Obj: 0.630907, No Obj: 0.530558, .5R: 0.400000, .75R: 0.000000, count: 5

Region 42 Avg IOU: 0.263569, Class: 0.696441, Obj: 0.464465, No Obj: 0.459764, .5R: 0.250000, .75R: 0.000000, count: 8

Region 54 Avg IOU: 0.149923, Class: 0.491762, Obj: 0.336601, No Obj: 0.467867, .5R: 0.000000, .75R: 0.000000, count: 30

2: 583.775513, 564.703674 avg, 0.000000 rate, 0.534891 seconds, 64 images

Loaded: 1.104566 seconds

so using nohup to collect the outputs:

nohup ./darknet detector train cfg/coco.data cfg/yolov3.cfg

and then:

cp nohup.out log/yolov3.log

then open the exe file and fill in the parameters:

The parameters you filled in is very important:

line: the lines of you log file

step: How many rows are sampled every time?

start: from which line

end: to which line

ignore: Ignore the value of the previous n times.

选择文件：choose you log file

newName: your extract file name

picName: Name of picture saved

so, in my log file, there is 489284 lines and after extract the file, there is only 29328 lines left.

I hope you enjoy using this software.