How to train with 4-channel png image with Tiny-Yolo?

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )

http://pjreddie.com/darknet/

Other

21.68k stars 7.96k forks source link

How to train with 4-channel png image with Tiny-Yolo? #4221

Open sourabhyadav opened 4 years ago

sourabhyadav commented 4 years ago

I have 4-channel RGBA png image. I wanted to train a tiny-yolo model from scratch. I made channels = 4 in my cfg file. But when I run the training it crashes after 1st iteration itself.

Crash log: Region Avg IOU: 0.000000, Class: nan, Obj: -nan, No Obj: -nan, Avg Recall: 0.000000, count: 8 Segmentation fault (core dumped)

Do we have any repo for training 4-channel input images?

Thanks for the help.

sdrll commented 4 years ago

I have currently the same issue