All my prediction is the same

[Amory0709]

Hello,
I tried to estimate my video. No matter what input data it takes, the output always thinks my fingers are fully curl. However they are no curl. Please help me!! Very urgent TT

[Amory0709]

New finding: the release of 0.1.1 uses [x,y,z] [] as input. But in your demo, you are using {x: number, y: number, z: number, name: string }[] as the estimator's input.