search

cvlab-epfl / LIFT

Code release for the ECCV 2016 paper
485 stars 168 forks source link

How to train the whole network with your provided Webcam and Oxford dataset? #15

Closed Childhoo closed 7 years ago

Childhoo commented 7 years ago

The tested code provided trained descriptor/detector/orientation network. However, I hope to train those network with the provided dataset. Under python-code/Utils/networks there are files like "eccv_combined.py" "eccv_base.py" and "network_base.py", which are used to build the network architectures. I am wondering if this is the start point to train the whole network in your paper? Could you please give me some brief hints how to train the network with your provided Webcam and Oxford dataset? Thank you very much in advance.

kmyi commented 7 years ago

Hi,

The training code is a mess at the moment. We will try to clean it up after the ICCV deadline and release it. In the meantime, you can look at some of the issues (maybe closed) related to implementing the training procedure if it helps.

Cheers, kwang