【CVPR 2020】图神经网络（GNN）论文和代码汇总

【导读】本届CVPR共收到了6656篇有效投稿，接收1470篇，其接受率在逐年下降，今年接受率仅为22%。本文为大家整理了CVPR 2020 图神经网络(GNN)相关的比较有意思的值得阅读的10篇论文，供大家参考—分别是：点云分析、视频描述生成、轨迹预测、场景图生成、视频理解、3D语义分割、指代表达式推理、图像描述生成、图像处理、时空图等方向文章。

1. Grid-GCN for Fast and Scalable Point Cloud Learning

• 论文链接：https://arxiv.org/abs/1912.02984

2. Object Relational Graph with Teacher-Recommended Learning for Video Captioning

• 论文链接：https://arxiv.org/abs/2002.11566

3. Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction

• 论文链接：https://arxiv.org/abs/2002.11927
• 代码地址：https://github.com/abduallahmohamed/Social-STGCNN

4. Unbiased Scene Graph Generation from Biased Training

• 论文链接：https://arxiv.org/abs/2002.11949
• 代码地址：https://github.com/KaihuaTang/Scene-Graph-Benchmark.pytorch

5. Where Does It Exist: Spatio-Temporal Video Grounding for Multi-Form Sentences

• 论文链接：https://arxiv.org/abs/2001.06891

6. Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions

• 论文链接：https://arxiv.org/pdf/2004.03967.pdf

7. Graph-Structured Referring Expression Reasoning in The Wild

• 论文链接：https://arxiv.org/pdf/2004.08814.pdf
• 代码地址：https://github.com/sibeiyang/sgmn

8. Say As Y ou Wish: Fine-grained Control of Image Caption Generation with Abstract Scene Graphs

• 论文链接：https://arxiv.org/pdf/2003.00387.pdf

9. Semantic Image Manipulation Using Scene Graphs

• 论文链接：https://www.researchgate.net/publication/340523427_Semantic_Image_Manipulation_Using_Scene_Graphs
• 代码地址：https://he-dhamo.github.io/SIMSG/

10. Spatio-Temporal Graph for Video Captioning with Knowledge Distillation

• 论文链接：https://arxiv.org/pdf/2003.13942.pdf