【论文速读，找找启发点】2024/6/16

ICME 2023

End-To-End Part-Level Action Parsing With Transformer

类似 DETR，通过 加 query的方式实现 端到端

ELAN: Enhancing Temporal Action Detection with Location Awareness

如何实现位置感知？ => 重叠的卷积核？

Do we really need temporal convolutions in action segmentation?

No, 用 Transformer

Explicit Attention Modeling for Pedestrian Attribute Recognition

任务：Pedestrian Attribute Recognition

A Simple Masked Autoencoder Paradigm for Point Cloud

点云，自编码器

RF-based Multi-view Pose Machine for Multi-Person 3D Pose Estimation

多人, 3D姿态估计

---

中国计算机学会推荐国际学术会议和期刊目录（2022）：https://ccf.atom.im/
搜索文献：dblp