Can not testing Pretrained Models. about mmskeleton HOT 3 CLOSED

Hi, @124399839,
It seems like you loaded the old model checkpoint in the new codebase. Please run bash tools/get_models.sh first.

from mmskeleton.

Hi@yysijie @yjxiong，Thanks for your great work!
我現在已經可以跑通測試程序（python main.py recognition -c config/st_gcn/kinetics-skeleton/test.yaml）。
現在在跑Demo程序時出現下面的錯誤，
deeplearning@server:~/code$ python3 main.py demo --openpose ~/project/openpose/build/ --video ./resource/media/skateboarding.mp4
Starting OpenPose demo...
Auto-detecting all available GPUs... Detected 1 GPU(s), using 1 of them starting at GPU 0.
Starting thread(s)...
OpenPose demo successfully finished. Total time: 17.060792 seconds.
Traceback (most recent call last):
File "main.py", line 31, in
p.start()
File "/home/dl/code/processor/demo.py", line 46, in start
video = utils.video.get_video_frames(self.arg.video)
File "/home/dl/code/tools/utils/video.py", line 35, in get_video_frames
vread = skvideo.io.vread(video_path)
File "/usr/local/lib/python3.6/dist-packages/skvideo/io/io.py", line 133, in vread
assert _HAS_FFMPEG, "Cannot find installation of real FFmpeg (which comes with ffprobe)."
AssertionError: Cannot find installation of real FFmpeg (which comes with ffprobe).
出現這個錯誤之後我做了如下檢查：首先單獨測試openpose.bin，是可以檢測人體關鍵骨骼點的。
然後查看data/openpose_estimation/snippets/skateboarding下生成的json文件，我查看每個json文件裏面的內容都是空的，並沒有檢測到關鍵骨骼點的座標，不知道產生錯誤的原因是否和這個有關係。
謝謝！

from mmskeleton.

@124399839, 谢谢你指出这个问题，
I think you should install FFmpeg first.
Try to run sudo apt-get install ffmpeg in your console.
Thank you for your report, and we will add this pre-requirement into README.

from mmskeleton.