search

ultralytics / yolov5

YOLOv5 🚀 in PyTorch > ONNX > CoreML > TFLite
https://docs.ultralytics.com
GNU Affero General Public License v3.0
51.24k stars 16.44k forks source link

Cls loss 0 #3064

Closed debparth closed 3 years ago

debparth commented 3 years ago

❔Question

I'm training Yolov5m model over my dataset which has only 1 class in it. While training I'm seeing that my cls loss is 0 in each and every epoch. Is it because of the reason that I only have 1 class in my dataset or due to some other reason?

Yassine-Fak commented 3 years ago

Hi! I think that this is due to the fact that you have only one class in your dataset. Is that right @glenn-jocher ? By the way @glenn-jocher thank you so much for this amazing great repository and job!!!!

glenn-jocher commented 3 years ago

@Yassine-Fak @debparth single-class datasets don't incur class loss. If I asked you to pick a number between 0 and 0, you wouldn't have to think hard would you?

debparth commented 3 years ago

@glenn-jocher yea... true. And one more thing while training the img_size differs to the param passed.

image

like here it shows that I want to use 864 then why does it shows 1728?

glenn-jocher commented 3 years ago

@debparth If you believe you have a reproducible issue, we suggest you close this issue and raise a new one using the 🐛 Bug Report template, providing screenshots and a minimum reproducible example to help us better understand and diagnose your problem. Thank you!