search

oeway / pytorch-deform-conv

PyTorch implementation of Deformable Convolution
MIT License
911 stars 151 forks source link

Why only fine tuning on deformable convnet #18

Closed Esaada closed 5 years ago

Esaada commented 5 years ago

Hi, why train first on the convnet and only after do finetuning on the deformable convnet? why not straight traing on deformable (with freezing the ofsets as written in the paper)

oeway commented 5 years ago

Hi, please refer to the original paper or the tensorflow implementation: https://github.com/felixlaumon/deform-conv This repo is just a port from that one.

Edit: it seems the original repo has been removed, but I think this one is the same: https://github.com/kastnerkyle/deform-conv