search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.66k stars 7.96k forks source link

Custom multi-channel input data processing #8003

Open zrion opened 3 years ago

zrion commented 3 years ago

Hello,

I'm working on a project where I uses YOLO to do detection on non-image data. Currently my approach to to force the input data to have image-like format (scaled pixel-valued 3 channels). To improve the performance, I would like to modify the model to support multi-channel data (maybe more than 3) with my custom preprocessing. Could you give me some pointers of how to integrate this feature in a most efficient way? Thank you!

mjack3 commented 3 years ago

Hello @zrion Actually i am working in the same, but in my case i need to use just 2 channels. Would be great share any information. At the moment today is my first day in the task, did you solve anything?

zrion commented 3 years ago

@AlexeyAB could you please provide some advice? Thank you!