The Android Example for "pose_landmarker" is not working for videos

[LeonelZalegas]

When you try to select a video that is longer than 10 seconds, it keeps loading forever.
I checked the code, and basically it takes the video, obtain all the frames at once and process every single of them with mediapipe (this is where the loading view appears in the UI), this is something that I guess requires a lot of processing and that's why it won't work with videos longer than 10 sec

Isn't possible to implement something like the implementation that is done for the camera but for the video analysis ? In the camera processing we are returning the frames in real time and immediately processing it with mediapipe and return it so it can be shown in the UI