search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.58k stars 7.94k forks source link

distinguish traffic signs #3739

Open AshleyRoth opened 4 years ago

AshleyRoth commented 4 years ago

hello! Is it possible to distinguish between traffic signs? Most interested in speed limit signs. Outwardly, they are the same, they differ only within the speed limit figures. I attach examples of the image.

in this example, can darknet distinguish between a speed limit of 50 and a speed limit of 70? How to properly train for this?

50 70

AlexeyAB commented 4 years ago

Yes.

for each object which you want to detect - there must be at least 1 similar object in the Training dataset with about the same: shape, side of object, relative size, angle of rotation, tilt, illumination. So desirable that your training dataset include images with objects at diffrent: scales, rotations, lightings, from different sides, on different backgrounds - you should preferably have 2000 different images for each class or more, and you should train 2000*classes iterations or more

Read more: https://github.com/AlexeyAB/darknet#how-to-train-to-detect-your-custom-objects

sctrueew commented 4 years ago

@AlexeyAB Hi,

I have same problem about the only speed limit and which parameters can I change to increase the accuracy? Can I change the hue or saturation? Is it effective?

Thanks