Hi Ravi Theja

We are also training YOLO to detect number plates.

We have our own dataset of about 300+ images of vehicles but once we start training YOLO we are encountering NaN values for various fields but not for avg field which appears only after every 100 iterations.

Is Training going well or not ? I am sharing snapshots of my terminal while training is going.

This is what my terminal looks like at starting :

mask_scale: Using default ‘1.000000’
Loading weights from darknet19_448.conv.23…Done!
Learning Rate: 0.001, Momentum: 0.9, Decay: 0.0005
Resizing
608
Loaded: 0.017689 seconds
Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.559301, Avg Recall: -nan, count: 0
1: 636.425964, 636.425964 avg, 0.000000 rate, 32.543504 seconds, 1 images
Loaded: 0.000097 seconds
Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.552539, Avg Recall: -nan, count: 0
2: 625.538574, 635.337219 avg, 0.000000 rate, 31.364448 seconds, 2 images
Loaded: 0.000037 seconds
Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.560160, Avg Recall: -nan, count: 0
3: 641.501038, 635.953613 avg, 0.000000 rate, 31.238904 seconds, 3 images
Loaded: 0.000043 seconds
Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.559215, Avg Recall: -nan, count: 0
4: 640.466675, 636.404907 avg, 0.000000 rate, 31.150311 seconds, 4 images
Loaded: 0.000047 seconds

This is what my terminal looks like after 300th iteration :

Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.559009, Avg Recall: -nan, count: 0
299: 253.360443, 370.132019 avg, 0.000008 rate, 11.988998 seconds, 299 images
Loaded: 0.000051 seconds
Region Avg IOU: -nan, Class: -nan, Obj: -nan, No Obj: 0.561851, Avg Recall: -nan, count: 0
300: 257.910461, 358.909851 avg, 0.000008 rate, 12.007875 seconds, 300 images
Saving weights to backup//yolov2-voc.backup
Saving weights to backup//yolov2-voc_300.weights
Resizing

I would really appreciate if you could please share the weights file that you got after training your YOLO to detect number plates.