shubhtuls / ViewpointsAndKeypoints

Code Release for "Viewpoints and Keypoints" CVPR 2015.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Update caffe ?

bermanmaxim opened this issue · comments

Hello,

I would like to train the networks here but I would need to use the iter_size options available on newer versions of caffe in order to make the training fit my limited GPU memory... Is there any chance you could either update your modified version of caffe or provide the trained network snapshots with the release for easier result reproduction ?

Thanks!

I think it might be easiest for me to release the trained models. The viewpoint prediction model (for all 3 angles) can be found here. Please let me know if you require the keypoint prediction model. If you would like to train the viewpoint prediction model on your end, I can work on releasing a cleaner python version of code but that might take longer.

Thanks for the trained model; that makes it much easier. If possible I would indeed be interested by the keypoint model as well. Best

There are two keypoint prediction models for different scales (6 X 6 and 12 X 12).

Allright ; thanks a lot for the trained Caffemodels.