How to load the models only once ?

MVIG-SJTU / AlphaPose

Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System

http://mvig.org/research/alphapose.html

Other

8.06k stars 1.98k forks source link

How to load the models only once ? #1050

Closed SachaCutillas closed 2 years ago

SachaCutillas commented 2 years ago

Hello !

First, thank you for your work, it's amazing. I woud like to implement AlphaPose in a real time project and I would like to load the SE Resnet, YOLO, pose and reid model only once and execute AlphaPose frame after frame. For now, I can just run the command python demo_inference.py and it loads the model each time I call it. How can I achieve that ?

Thank you vor your time !

Fang-Haoshu commented 2 years ago

Hi, for real-time stream, you can use the flag --webcam $webcam_id(for example, 0) You can refer to our document mentioned in the readme. We have provided many document.

Fang-Haoshu commented 2 years ago

More detailed inference options and examples, please refer to GETTING_STARTED.md