Process real-time video streams

Vegetebird / StridedTransformer-Pose3D

[TMM 2022] Exploiting Temporal Contexts with Strided Transformer for 3D Human Pose Estimation

MIT License

335 stars 37 forks source link

Closed Allencheng97 closed 2 years ago

Allencheng97 commented 2 years ago

HI, I saw your answer in a closed issue. May I ask how can I retrained in casual setting to process real-time video streams?

Vegetebird commented 2 years ago

You can supervise the output 3D pose with the end-of-frame GT 3D pose rather than the center-frame GT 3D pose.

Terrywang2001 commented 4 months ago

Hi, did you manage to retrain in causal setting?