百度飞桨论文复现营

课程地址
https://aistudio.baidu.com/aistudio/education/group/info/1340

这次的论文复现营，时长上比以往的训练营时间长。复现论文是一项理论加实际的活，首先要理解意思，还要有编程基础。这是一个挑战，如果这个成功坚持下来了，就是一项了不起的事情。这次有两个方向，一个是gan，还有一个是视频分类，都可供选择。两个方向都分别会有5篇论文带读。要求用paddle框架复现。
GAN方向的论文
1.LARGE SCALE GAN TRAINING FOR HIGH FIDELITY NATURAL IMAGE SYNTHESIS
https://github.com/sxhxliang/BigGAN-pytorch
2. Few-shot Video-toVideo Synthesis
https://github.com /NVlabs/few-shotvid2vid
3. First Order Motion Model for Image Animation
https://github.com/AliaksandrSiarohin/first-order-model
4.StarGAN v2: Diverse Image Synthesis for Multiple Domains
https://github.com/clovaai/stargan-v2
5. U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation
https://github.com/znxlwm/UGATIT-pytorch
视频分类4篇论文
1.ECO: Efficient Convolutional Network for Online Video Understanding
网址：https://github.com/mzolfaghari/ECO-pytorch
2.Temporal Pyramid Network for Action Recognition
网址：https://github.com/decisionforce/TPN
3.Learning Spatio-Temporal Features with 3D Residual Networks For Action Recognition
网址：https://github.com/kenshohara/3D-ResNets-PyTorch
4.Representation Flow for Action Recognition
网址：https://github.com/piergiaj/representation-flow-cvpr19