[OTA] Optimal Transport Assignment
{, }
Paper: https://arxiv.org/abs/2103.14259
Code:
Embedded Files
Motivation, Objectives and Related Works
Motivation, Objectives and Related Works
Motivation
Motivation
Objectives
Objectives
A method that dynamically matches correct labels and prediction results depending on the image situation. The Sinkhorn-Knopp algorithm is used to calculate the optimal transportation cost from the transportation source (correct label) to the transportation destination (prediction result).
In OTA, the number of prediction results assigned to each correct label is changed by considering the size and overlap of the correct labels (dynamic top-k strategy).
Anchors that are ambiguous as to which correct label they belong to can also be assigned.
Related Works
Related Works
Methods
Methods
Method 2
Method 2
Experimental Results
Experimental Results
Experimental Results
Experimental Results
Ablations
Ablations
References
References
Ge, Zheng, et al. "Ota: Optimal transport assignment for object detection." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021.
n2 n0
θ
Page updated
Google Sites
Report abuse