WongKinYiu / yolov9

Implementation of paper - YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information
GNU General Public License v3.0
8.94k stars 1.41k forks source link

Training with a npy file #449

Open DeepSkyBlue123 opened 5 months ago

DeepSkyBlue123 commented 5 months ago

So i am currently trying to train with a npy file(An image with 5 channels), but it does not work as the file format is not accepted. How do i fix this. Also do i need to train from scratch as the models and pretrained weight are trained using images with 3 channels?

WongKinYiu commented 4 months ago

You have to modify the dataloader. For loading .npy files as training data, you could take a look --cache disk. But the data augmentation methods are almost only support 3 channels inputs, you have to modify them.

DeepSkyBlue123 commented 4 months ago

Thank you! I was able to do it. However i had to comment out some of the image augmentation methods as it was hard coded to be for 3 channels