As illustrated in paper, bilinear sampling is used for the feature maps of the corresponding proposals. However, in the demo.py or validation.py, I found it directly rotates the proposals into horizontal direction and then feed them into recognition network. So, I would like to know where is the bilinear sampling applied?
Hey,
As illustrated in paper, bilinear sampling is used for the feature maps of the corresponding proposals. However, in the demo.py or validation.py, I found it directly rotates the proposals into horizontal direction and then feed them into recognition network. So, I would like to know where is the bilinear sampling applied?
Thanks and have a nice day.