Closed technicalReport closed 7 years ago
Check https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/doc/installation.md#faq In particular, the Q: Low speed
question. Post again if you have any further question about it (and please also post your hardware if you post again, so we can better answer). Thanks
Thanks for your kindly reply. Now I try it with the command ' ./build/examples/openpose/openpose.bin --video ../zhenduoyan.mp4 --net_resolution 320x176 --model_pose MPI_4_layers --process_real_time --num_gpu 4' , the average fps turns to be 1.6, which is still much slower than the result mentioned in the paper.
BTW: I test it in the remote linux server and the output video sends back via xming.
Below is my server configuration:
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.51 Driver Version: 375.51 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+=======|
| 0 Tesla K80 Off | 0000:06:00.0 Off | 0 |
| N/A 56C P0 66W / 149W | 1154MiB / 11439MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla K80 Off | 0000:07:00.0 Off | 0 |
| N/A 45C P0 77W / 149W | 956MiB / 11439MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 2 Tesla K80 Off | 0000:84:00.0 Off | 0 |
| N/A 62C P0 66W / 149W | 956MiB / 11439MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 3 Tesla K80 Off | 0000:85:00.0 Off | 0 |
| N/A 48C P0 124W / 149W | 956MiB / 11439MiB | 89% Default |
If you are sending back the output video the speed is highly conditioned by that. Try without any visual output (--no_render_pose
) to see the real speed.
EDITED: closed for now. Please post again if you have any other related question.
Issue summary
I have some question about its real time speed. I have run it with some dataset, but its performance is not fast enough to be 'real time'.