About the loss

Question

About the loss

door5719 opened this issue 6 years ago · 4 comments

About the loss

Answer 1 · 2018-07-11T14:43:44.000Z

Hi @door5719 .I think you want to implement that, but in fact I haven't finished reading this paper. I will share my idea( i hope implement) after I study it.. Thanks for sharing good paper.

Answer 2 · 2018-07-12T00:59:47.000Z

I'm very glad to get your reply.If you finish implementing that paper,please give me some advice.Thank you very much.

Answer 3 · 2018-07-15T17:54:10.000Z

Hi @door5719 ,
Truth means Ground Truth(GT). We have to compare with GT to get the loss of the prediction box, but there are many prediction boxes and GT's on the image. So we use IOU to see which predictive boxes and GT are responsive.

Answer 4 · 2018-07-16T03:15:52.000Z

@springkim thank you very much