DenisTome
commented
6 years ago Hi, Yes I know. In our case we had a cython optimized version of that code that is much more efficient, saving in the cache some matrices for saving time.
On Aug 21, 2018, at 1:13 AM, GreatGBL notifications@github.com wrote:
Greeting! I am running the 3D pose by this paper (Lifting from the deep), the 2D human joints are from Oppenpose (https://github.com/ildoonet/tf-pose-estimation)
I test each frame I need approximate 0.05s to process the 2D joints, while it requires near 10s to transform 3D, so it is impossible to construct real-time system. However, I see the demo in Lifting, it is real-time, how do you do that? And may I ask how to speed up the 3D transform ? Thanks!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
dandingol03
Greeting! I am running the 3D pose by this paper (Lifting from the deep), the 2D human joints are from Oppenpose (https://github.com/ildoonet/tf-pose-estimation)
I test each frame I need approximate 0.05s to process the 2D joints, while it requires near 10s to transform 3D, so it is impossible to construct real-time system. However, I see the demo in Lifting, it is real-time, how do you do that? And may I ask how to speed up the 3D transform ? Thanks!