xuannianz / keras-GaussianYOLOv3

Gaussian YOLOv3 (An Accurate and Fast Object Detector Using Localization Uncertainty for Autonomous Driving (ICCV, 2019)) implementation in Keras and Tensorflow

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Can we try Gaussian Method on yolo_v2

Daniel-1597 opened this issue · comments

Can you explain to me the lines below what is the use of these lines?

fm_13_input = Input(shape=(None, None, num_anchors_per_layer, num_classes + 5), name='fm_13_input')

fm_26_input = Input(shape=(None, None, num_anchors_per_layer, num_classes + 5), name='fm_26_input')
fm_52_input = Input(shape=(None, None, num_anchors_per_layer, num_classes + 5), name='fm_52_input')
  1. I think we can. They use the same logic to predict bounding boxes.
  2. There three inputs would be assigned to ground truth of the feature maps so that we can compute the loss conveniently.

1905.01296v2.pdf
If possible can you implement this paper

Aha, it's indeed beyond my knowledge.

That's the toughest paper I have ever read

commented

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.