search

AlvinYH / Faster-VoxelPose

Official implementation of Faster VoxelPose: Real-time 3D Human Pose Estimation by Orthographic Projection
MIT License
156 stars 18 forks source link

the bounding box isn't sufficiently large to cover poses! #25

Closed FrankMMMMMM closed 1 year ago

FrankMMMMMM commented 1 year ago

Thanks for your work！When I run "python run/train.py --cfg configs/campus/jln64.yaml". I meet the following problem: image Can you help me to solve this problem?

1124792823 commented 1 year ago

same question

FrankMMMMMM commented 1 year ago

same question

@1124792823 Hi That doesn't seem to be a problem.

JiatengLiu commented 1 year ago

Hi! How did you solve this problem?

FrankMMMMMM commented 1 year ago

@CH1-pix I didn't solve it. But in fact you can ignore the problem. The experiment seems to work fine. Killed may be due to lack of computer performance.

By the way, did you find any other code that does multi-camera target tracking?

JiatengLiu commented 1 year ago

@FrankMMMMMM Thanks!! On what GPUs did you complete the training? I can't do it with the RTX 3050 LAPTOP. Regarding your question, I remember that CVPR2023 has a related paper, you can go and take a look.

FrankMMMMMM commented 1 year ago

@CH1-pix I trained on the Campus dataset using an RTX2080TI in a cloud GPU environment.