search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.65k stars 7.96k forks source link

Invalid SOS parameters for sequential JPEG #3312

Open ReynardTing opened 5 years ago

ReynardTing commented 5 years ago

i have added training dataset but this error appeared while training Screenshot from 2019-06-04 10-30-00

ReynardTing commented 5 years ago

what does this error imply?

ReynardTing commented 5 years ago

does it affect with my training if the avg losses did not show nan

AlexeyAB commented 5 years ago

It seems it is just a warning, not an error. So it shouldn't affect on trainig.

ReynardTing commented 5 years ago

what should the avg loss(mine was around 0.7) at that point of the training. Cause when i run the ./darknet detector demo data/obj.data yolov3-tiny_3l.cfg yolov3-tiny_3l_20000.weights no bounding boxes showed up on my webcam for the trained object what does this mean? thanks :)

AlexeyAB commented 5 years ago

what mAP do you get? https://github.com/AlexeyAB/darknet#when-should-i-stop-training

ReynardTing commented 5 years ago

mine was 0.7 and it still can decrease i think i was trying testing it to see if it would work cause is it normal to not show any bounding boxes at all?