corner loss misuse label info?
windyrobin opened this issue · 1 comments
windyrobin commented
heading = angle_decode(heading_res_norm, heading_class_label)
size = size_decode(size_res_norm, mean_size_array, size_class_label)
corners_loss, corner_gts = self.get_corner_loss(
(center_preds, heading, size),
(center_label, heading_label, size_label)
)
heading
use heading_class_label
as the main part, should use heading_pred
instead?
zhixinwang commented
Regularization Loss. Use the heading class label to avoid the accumulative error especially at the early stage.