Yang Zhou
Yang Zhou
【队名】:qifei 【序号】:40 【论文】:[HINet: Half Instance Normalization Network for Image Restoration](https://arxiv.org/pdf/2105.06086v1.pdf) 【状态】:报名&提交 【repo链接】https://github.com/youngAt19/hinet_paddle
@vincentfung13 Hi, in your paper you said that the model was trained on the 20 city sequences from the KITTI Raw dataset, and evaluated on another 4 city sequences. Besides,...
@vincentfung13 Wow! Thanks a lot!!
I think actually the author is using w space. The discriminator is implemented as a component in encoder. And the two returning terms of model.encode is w and d's output...
Thanks for your reply. I thought it just takes you 53 hours to train the model from [here](https://github.com/CoinCheung/DenseCL#:~:text=using%20this%20setting.-,Linear%20classification,-results%20on%20ImageNet)
Thanks for your reply. I thought it just takes you 53 hours to train the model from [here](https://github.com/CoinCheung/DenseCL#:~:text=using%20this%20setting.-,Linear%20classification,-results%20on%20ImageNet)