def __call__(self, ds: Detections, ctx: FrameContext) -> torch.Tensor
Return a constant-filled buffer shaped for the given detections batch.
Source: unitrack/states/identity.py:225
