search

bertinetto / siamese-fc

Arbitrary object tracking at 50-100 FPS with Fully Convolutional Siamese networks.
http://www.robots.ox.ac.uk/~luca/siamese-fc.html
MIT License
619 stars 224 forks source link

why use the totalStride? #57

Open brealisty opened 6 years ago

brealisty commented 6 years ago

I don't understand the code disp_instanceInput = disp_instanceFinal * p.totalStride / p.responseUp;

why *p.totalStride. cuz the stride of conv?

jvlmdr commented 6 years ago

Yes. Usually there will be several layers in the network with stride > 1. This means that displacements in the response map need to be multiplied by the overall stride of the network.

guwenning commented 5 years ago

@jvlmdr Hello, I don't quite understand the 8 coordinates in ground_truth. I used a trace with a number of coordinates of 4. I found that the coordinates of the first frame did not completely wrap the target on the target. If I use my own sequence of images, I should enter the first What should be the coordinate format of a frame of picture?