Darknet: No Bounding box coordinates

[spandanji]

if(bot > im.h-1) bot = im.h-1; printf("%s: %.0f%%", names[class_id], prob * 100); printf("\n - id: %d, x_start: %d, y_start: %d, x_end: %d, y_end: %d", class_id, left, top, right , bot);
I did the changes and then I did make clean followed by make. But the changes are not reflected. I still cannot see the bounding boxes. I had trained the model for single class classification prior to this. Do I need to retrain the model just to get the coordinates?

[spandanji]

I also used a printf statement in image_opencv.cpp but it had no effect. I am unable to get the coordinates

[mehio]

Any luck???
I did the same! Still nothing!
I noticed that it works for people with pjreddies image.c. The image.c in alexeyAB is different and this modification you are talking about is not working.

[mehio]

@another-dark-knight Let me know please

[spandanji]

@mehio No luck on the .c file. I used the python API to print out the coordinates.

[mehio]

Hey i managed to print the coordinates using a darknet command . No need to use the .c file .

./darknet detector test cfg/voc.data yolo-voc.cfg yolo-voc.weights -dont_show -ext_output < data/train.txt > result.txt

Im using alexeyAB fork, not pjreddies

[shachitaku]

https://github.com/AlexeyAB/darknet#how-to-use-on-the-command-line
Output coordinates of objects: darknet.exe detector test cfg/coco.data yolov3.cfg yolov3.weights -ext_output dog.jpg

is this not the answer?
when i do it on mine on terminal i get things that look like this:
test2.png: Predicted in 11.823000 milli-seconds.
Kanji_Ten: 81% (left_x: 2 top_y: 181 width: 10 height: 13)
Kanji_Ten: 90% (left_x: 94 top_y: 398 width: 11 height: 10)
Kanji_Ten: 99% (left_x: 272 top_y: 430 width: 12 height: 12)
Kanji_Ten: 94% (left_x: 23 top_y: 633 width: 17 height: 18)
Kanji_Ten: 98% (left_x: 371 top_y: 234 width: 10 height: 18)
Kanji_Ten: 60% (left_x: 432 top_y: 410 width: 16 height: 17)
Kanji_Ten: 100% (left_x: 700 top_y: 558 width: 21 height: 21)
Kanji_Ten: 100% (left_x: 719 top_y: 240 width: 11 height: 12)

@mehio thanks i could output to .txt file using your way :)