π₯CDIoU and CDIoU loss is like a convenient plug-in that can be used in multiple models. CDIoU and CDIoU loss have different excellent performances in several models such as Faster R-CNN, YOLOv4, RetinaNet and . There is a maximum AP improvement of 1.9% and an average AP of 0.8% improvement on MS COCO dataset, compared to traditional evaluation-feedback modules. Here we just use as an example to illustrate the code.
Thank you for your work! I'm very interested in it.
In your paper, the step 4 in algorithm of loss function is shown as follows:
4: compute L CDIoU = L IoUs + diou, L IoUs could be
L IoU = β ln(IoU ), L IoU = 1β IoU , L IoU = 1β IoU
or L DIoU , L CIoU ;
Do you think whether the bold part is wrong? Is it L GIoU instead?
BTW, whether the CDIoU loss could be another format? For example:
L CDIoU = 1 - CDIoU or L CDIoU = - ln(CDIoU)
Thank you for your work! I'm very interested in it. In your paper, the step 4 in algorithm of loss function is shown as follows:
4: compute L CDIoU = L IoUs + diou, L IoUs could be L IoU = β ln(IoU ), L IoU = 1β IoU , L IoU = 1β IoU or L DIoU , L CIoU ;
Do you think whether the bold part is wrong? Is it
L GIoU
instead? BTW, whether the CDIoU loss could be another format? For example:L CDIoU = 1 - CDIoU
orL CDIoU = - ln(CDIoU)