The raspberry pies were not quick enough for our earlier implementation of image recogntion.
WHAT?
I implemented new versions of pose landmarker and gesture recognizer based on the earlier ones, but with the option to choose when results are applied to the images. This allows the video analysis node to only apply the results every e.g. 10 frames. Relevant changes have also been made to the video_analyzer node.
WHY?
The raspberry pies were not quick enough for our earlier implementation of image recogntion.
WHAT?
I implemented new versions of pose landmarker and gesture recognizer based on the earlier ones, but with the option to choose when results are applied to the images. This allows the video analysis node to only apply the results every e.g. 10 frames. Relevant changes have also been made to the video_analyzer node.