Luca Bertinetto

Hi, For some reason, the problem only occurred when using PyTorch 1.4. Using PyTorch 1.0 results are reproduced. Will post here when/if I can find a fix for PyTorch 1.4

Hi, I got the same as @csyanbin with python 3.5, cuda 8 and pytorch 0.3

Hi, At inference time, the response is upsample. This has to be taken into account when computing the position in original frame coordinates. Best, Luca On Tue, Jul 9, 2019...

Hi, Thanks for pointing it out. On a first look, your script seems fine, but to be honest I have some troubles remembering the exact setup, as more than one...

Yes, that link is still active. We are having problems with the FTP. I am updating the page. Thank you. On Sun, Mar 11, 2018 at 1:30 AM, Dat Do...

Hi, It is used to dampen the scale factor, so that abrupt changes in scale do not have distruptive effects.

Hi, Definitely. Thanks!

Hi, Do you mean this Google Drive link? https://drive.google.com/file/d/0B7Awq_aAemXQZ3JTc2l6TTZlQVE/view It seems fine to me. Do you have problems with Drive in general or just for this file? On Mon, Dec...

This version uses a matconvnet-pretrained network. You can train new models using the code from the [conv5 baseline of CFNet](https://github.com/bertinetto/cfnet/blob/master/src/training/run_experiment_baseline_conv5.m) However, since the model is fully specified in this TensorFlow...

I have actually finished with the pytorch version a couple of weeks ago but it's a bit behind the original implementation performance. I might try and release it, perhaps people...