kkhoot / PAA

A PyTorch implementation of the paper `Probabilistic Anchor Assignment with IoU Prediction for Object Detection` ECCV 2020 (https://arxiv.org/abs/2007.08103)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

locations

qianwangn opened this issue · comments

That parameter seems not used.

def compute_reg_loss(self, regression_targets, box_regression, all_anchors, labels, locations, n_anchors, weights, N): if 'iou' in self.reg_loss_type: reg_loss = self.GIoULoss(box_regression, regression_targets, all_anchors, weight=weights) elif self.reg_loss_type == 'smoothl1': reg_loss = smooth_l1_loss(box_regression, regression_targets, beta=self.bbox_reg_beta, size_average=False, sum=False) if weights is not None: reg_loss *= weights else: raise NotImplementedError return reg_loss[labels > 0].view(-1)

Hi, thanks for reporting. I should have used some linters before releasing the code. :( I will clean this up soon. Thanks!

closing this as it is fixed now. Thanks again!