Shu Kong

Thanks for your interest. I've added demo1 with all useful script. I'll add in other demos in the next a few days.

Hi, Thanks for your interest. I would release that after our submission becomes accepted. But here is how we did that -- 1. prepare 15 maps for 15 keypoints respectively....

I just simply used scipy.ndimage.binary_dilation to dilate each joint before propagating; then as for voting, I compute the coordinate mean (based on mass center) of the the propagated dots. Regards,...

Hi, the error explains that the input mean has different size from the one required in the proto file. I guess one remedy is to crop the input mean into...

Hi, as I didn't use python interface, I'd suggest compiling the provided customCaffe which has new layers. When you compile customCaffe, you only need to do it using the following....

I'm afraid you have to compile the customCaffe to run it. But before that, I want to make sure that you can run the demo in matlab ( https://github.com/aimerykong/deepImageAestheticsAnalysis/tree/master/demoRatingImages )...

Thanks for noticing me that! On Thu, Aug 17, 2017 at 11:23 PM, Yanbin Liu wrote: > According to the Matlab code in: https://github.com/aimerykong/ > deepImageAestheticsAnalysis/blob/master/demoRatingImages/ > main_demo4_suggestion.m#L52 > It...

Hi, thanks for your interest. You are right! The "cls" is the K-way classifier for the closed-set classification. OpenGAN repurposes the off-the-shelf features and learns a discriminator for open-set detection....

Yes, adding open training samples is much better. Please refer to the second last cell for how to do so ( https://github.com/aimerykong/OpenGAN/blob/main/demo_OpenSetSegmentation_training.ipynb ) When you say "the same epochs models...

hi, you can think of (b) ansd (c) as a branch of (a) On Mon, Jun 11, 2018, 01:09 ranjiewen wrote: > [image: 20180611160545] > > > - in this...