同步操作将从 Gitee 极速下载/AlphaPose 强制同步,此操作会覆盖自 Fork 仓库以来所做的任何修改,且无法恢复!!!
确定后同步将在后台操作,完成时将刷新页面,请耐心等待。
Q: - I can't open the webcam or video file.
A: Try re-install opencv-python
with version >= 3.3.1.11 by
pip3 uninstall opencv_python
pip3 install opencv_python --user
Many people meet this problem at https://github.com/opencv/opencv/issues/8471. The solution I use is
sudo cp <path to opencv source repo>/build/lib/python3/cv2.cpython-35m-x86_64-linux-gnu.so /usr/local/lib/python3.5/dist-packages/cv2/cv2.cpython-35m-x86_64-linux-gnu.so
The idea is to replace the cv2.so library provided by pypi with the one compiled from sources. You can check for more info at https://github.com/opencv/opencv/issues/8471.
Q: - I can't open the webcam with the latest opencv-python
A: Check if your device is valid by
ls /dev/video*
Usually you can find video0
, but if you have a device with other index like video3
, you can run the program by
python3 webcam_demo.py --webcam 3 --outdir examples/res --vis
Q1: - I meet Killed
when processing heavy task, like large videos or images with crowded persons.
A: Your system meets out of cpu memory and kills the program autoly. Please reduce the length of result buffer by setting the --qsize
flag. By default length, free cpu memory over 70G+ is recommended in heavy task.
Q2: - I meet segmentation fault when processing heavy task, like large videos or images with crowded persons.
A: The parallelization module torch.multiprocessing
is prone to shared memory leaks. Its garbage collection mechanism torch_shm_manager
may cause segmentation fault under long-time heavy load. We found this issue when processing large videos with hundreds of persons. To avoid this issue, you can set --sp
flag to use multi-thread instead, which sacrifices a little efficiency for more stablity.
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。