Pose for Everything: Towards Category-Agnostic Pose Estimation

估计一切姿态：面向类别无关的姿态估计.

Localization with Sampling-Argmax

通过Sampling-Argmax来定位.

Poseur: Direct Human Pose Regression with Transformers

Poseur：使用Transformer直接进行人体姿态回归.

Next-Generation Pose Detection with MoveNet and TensorFlow

通过MoveNet和TensorFlow实现下一代姿态检测.

Low-resolution Human Pose Estimation

低分辨率人体姿态估计.

TOOD: Task-aligned One-stage Object Detection

TOOD：任务对齐的单阶段目标检测方法.